Using Data Mining to Model Student Success 
 
by 
 
Becky Geltz 
 
Submitted in Partial Fulfillment of the Requirements 
for the Degree of 
Master in Computing and Information Systems 
in the 
Computing and Information Systems 
Program 
 
Youngstown State University 
December 2009 
 
 
 
 
 
 
Using Data Mining to Model Student Success 
 
Becky Geltz 
 
 
I hereby release this thesis to the public.  I understand that this thesis will be made available from 
the OhioLINK ETD Center and the Maag Library Circulation Desk for public access.  I also 
authorize the University or other individuals to make copies of this thesis as needed for scholarly 
research. 
 
Signature: 
     
  Becky Geltz, Student  Date 
 
Approvals: 
      
  John R. Sullins, Thesis Advisor   Date 
 
      
  Thomas A. Bodnovich, Committee Member Date 
 
      
  Robert A. Hogue, Committee Member  Date 
 
      
  Peter J. Kasvinsky, Dean of School of Graduate Studies and Research Date 
 
 
 
 
TABLE OF CONTENTS 
 
 
Abstract ............................................................................................................................................ i 
Acknowledgements ......................................................................................................................... ii 
 
Chapters 
1.  Motivation and Goals  .............................................................................................................1 
1.1 Motivation  .....................................................................................................................1 
1.2 Goals  .............................................................................................................................4 
2.  Data Collection and Processing  ..............................................................................................6 
2.1 Data Collection  .............................................................................................................6 
2.2 Data Processing  .............................................................................................................8 
2.3 The Cohort .....................................................................................................................8 
2.4 Data Mining Tool - Weka ............................................................................................33 
3.  Analysis .................................................................................................................................38 
4.  Discussion .............................................................................................................................66 
5.  Conclusions ...........................................................................................................................66 
5.1 Recommendations ........................................................................................................67 
References ..................................................................................................................................68 
Weka Run results ....................................................................................................... Appendix A 
 
Figures 
The Cohort 
Figure 1 – Gender ..............................................................................................................10 
Figure 2 – Age Groups .......................................................................................................10 
Figure 3 – Racial/Ethnic Background ................................................................................11 
Figure 4 – State Residency ................................................................................................11 
 
 
Figure 5 – Housing Status ..................................................................................................12 
Figure 6 – ACT Composite Score ......................................................................................12 
Figure 7 – High School Graduating GPAs ........................................................................13 
Figure 8 – Advance Placement Credits ..............................................................................13 
Figure 9 – Academic Intention ..........................................................................................14 
Figure 10 – Major Field of Study – First Term .................................................................15 
Figure 11 – Marital Status..................................................................................................15 
Figure 12 – Financially Dependent Upon Parents .............................................................16 
Figure 13 – Cost of Attendance .........................................................................................17 
Figure 14 – 9-Month Expected Family Contribution ........................................................18 
Figure 15 – Need Level......................................................................................................19 
Figure 16 – Received Any Financial Aid ..........................................................................20 
Figure 17 – Federal Financial Aid .....................................................................................20 
Figure 18 – State Aid .........................................................................................................21 
Figure 19 – Federal Work Study Aid .................................................................................21 
Figure 20 – Institutional Aid ..............................................................................................22 
Figure 21 – Other Third Party Aid .....................................................................................23 
Figure 22 – Student Loans .................................................................................................23 
Figure 23 – First Term Academic Load .............................................................................24 
Figure 24 – First Term Attempted Credit Hours ...............................................................24 
Figure 25 – Credit Hours Earned .......................................................................................26 
Figure 26 – First Term Total Quality Points Earned .........................................................27 
Figure 27 – First Term GPAs.............................................................................................28 
Figure 28 – Any Remediation ............................................................................................28 
Figure 29 – Remedial English ............................................................................................29 
Figure 30 – Remedial Math ...............................................................................................29 
Figure 31 – Reading & Study Skills ..................................................................................30 
Figure 32 – Center for Student Progress ............................................................................31 
Figure 33 – Returned Spring Term ....................................................................................31 
Figure 34 – Returned Spring and Next Fall .......................................................................32 
Figure 35 – Returned the Next Fall....................................................................................32 
 
 
Weka 
Figure 36 – GUI Chooser ...................................................................................................34 
Figure 37 – Explorer Interface ...........................................................................................35 
Figure 38 – Classify ...........................................................................................................36 
Figure 39 – Evaluation Options .........................................................................................37 
Figure 40 – Classifier Output .............................................................................................38 
Analysis 
Figure 41 – J48’s Success Rate ..........................................................................................39 
Figure 42 – Simplified Decision Tree ................................................................................44 
Figure 43 – Predicted Outcome .........................................................................................45 
Figure 44 – Actual Outcome ..............................................................................................45 
Figure 45 – J48’s Predictions with Actual Outcome .........................................................45 
Figure 46 – Actual Outcome with J48’s Prediction ...........................................................45 
Figure 47 – Students Earning a Bachelor Degree  .............................................................46 
Figure 48 – Gender Prediction ...........................................................................................47 
Figure 49 – Gender Outcome .............................................................................................47 
Figure 50 – Age Groups Prediction ...................................................................................47 
Figure 51 – Age Groups Outcome .....................................................................................47 
Figure 52 – Racial/Ethnic Background Prediction ............................................................48 
Figure 53 – Racial/Ethnic Background Outcome ..............................................................48 
Figure 54 – State Residency Prediction .............................................................................48 
Figure 55 – State Residency Outcome ...............................................................................48 
Figure 56 – Housing Status Prediction ..............................................................................49 
Figure 57 – Housing Status Outcome ................................................................................49 
Figure 58 – ACT Composite Score Prediction ..................................................................49 
Figure 59 – ACT Composite Score Outcome ....................................................................49 
Figure 60 – High School Graduating GPAs Prediction .....................................................50 
Figure 61 – High School Graduating GPAs Outcome .......................................................50 
Figure 62 – Advanced Placement Credits Prediction ........................................................50 
Figure 63 – Advanced Placement Credits Outcome  .........................................................50 
Figure 64 – Academic Intention Prediction .......................................................................51 
 
 
Figure 65 – Academic Intention Outcome  ........................................................................51 
Figure 66 – Major Field of Study – First Term Prediction ................................................52 
Figure 67 – Major Field of Study – First Term Outcome  .................................................52 
Figure 68 – Marital Status Prediction ................................................................................53 
Figure 69 – Marital Status Outcome  .................................................................................53 
Figure 70 – Financially Dependent Upon Parents Prediction ............................................53 
Figure 71 – Financially Dependent Upon Parents Outcome .............................................53 
Figure 72 – Cost of Attendance Prediction ........................................................................54 
Figure 73 – Cost of Attendance Outcome..........................................................................54 
Figure 74 – 9-Month Expected Family Contribution Prediction .......................................54 
Figure 75 – 9-Month Expected Family Contribution Outcome .........................................54 
Figure 76 – Need Level Prediction ....................................................................................55 
Figure 77 – Need Level Outcome ......................................................................................55 
Figure 78 – Received Any Aid Prediction .........................................................................55 
Figure 79 – Received Any Aid Outcome ...........................................................................55 
Figure 80 – Federal Financial Aid Prediction ....................................................................56 
Figure 81 – Federal Financial Aid Outcome......................................................................56 
Figure 82 – State Aid Prediction ........................................................................................56 
Figure 83 – State Aid Outcome .........................................................................................56 
Figure 84 – Federal Work Study Predication ....................................................................57 
Figure 85 – Federal Work Study Outcome ........................................................................57 
Figure 86 – Institutional Aid Prediction ............................................................................57 
Figure 87 – Institutional Aid Outcome ..............................................................................57 
Figure 88 – Other Third Party Aid Prediction ...................................................................58 
Figure 89 – Other Third Party Aid Outcome .....................................................................58 
Figure 90 – Student Loan Prediction .................................................................................58 
Figure 91 – Student Loan Outcome ...................................................................................58 
Figure 92 – First Term Academic Load Prediction ...........................................................59 
Figure 93 – First Term Academic Load Outcome .............................................................59 
Figure 94 – First Term Attempted Credit Hours Prediction ..............................................59 
Figure 95 – First Term Attempted Credit Hours Outcome  ...............................................59 
 
 
Figure 96 – Credit Hours Prediction ..................................................................................60 
Figure 97 – Credit Hours Outcome  ...................................................................................60 
Figure 98 – First Term Total Quality Points Earned Prediction ........................................60 
Figure 99 – First Term Total Quality Points Earned Outcome ..........................................60 
Figure 100 – First Term GPA Prediction ...........................................................................61 
Figure 101 – First Term GPA Outcome ............................................................................61 
Figure 102 – Any Remediation Prediction ........................................................................61 
Figure 103 – Any Remediation Outcome  .........................................................................61 
Figure 104 – Remedial English Prediction ........................................................................62 
Figure 105 – Remedial English Outcome  .........................................................................62 
Figure 106 – Remedial Mathematics Prediction ................................................................62 
Figure 107 – Remedial Mathematics Outcome .................................................................62 
Figure 108 – Reading & Study Skills Prediction ...............................................................63 
Figure 109 – Reading & Study Skills Outcome.................................................................63 
Figure 110 – Center for Student Progress Prediction ........................................................63 
Figure 111 – Center for Student Progress Outcome ..........................................................63 
Figure 112 – Returned Spring Term Prediction .................................................................64 
Figure 113 – Returned Spring Term Outcome ..................................................................64 
Figure 114 – Returned Spring and Next Fall Prediction....................................................64 
Figure 115 – Returned Spring and Next Fall Outcome .....................................................64 
Figure 116 – Returned Next Fall Prediction ......................................................................66 
Figure 117 – Returned Next Fall Outcome ........................................................................66 
 
Tables 
Table 1 – Prediction and Outcome Distributions  ...................................................... Appendix B 
i 
 
Abstract 
As funding for higher education through federal and state sources continues to 
decline, and a stronger call for accountability is placed upon higher education institutions 
to graduate students within the expected amount of time, colleges and universities are 
looking for ways to best leverage their resources to attract college-ready students who 
will enroll in their institutions, remain enrolled consistently, and earn their undergraduate 
degrees in a timely manner.  Federal research conducted by the U.S. Department of 
Education’s National Center for Education Statistics through the Integrated 
Postsecondary Education Data System (IPEDS) examines aggregate student enrollment, 
degree completions, and graduation rates.  But to be truly helpful to the institutional 
researcher, unit record data is required.  Only by examining the many attributes of each 
individual student can an institution determine the unique characteristics which will lead 
to student academic success – degree attainment.  Because of the overall readability and 
the strong level of accuracy they can produce, decision trees are a good method for 
identifying the relationships between attributes in large datasets.  Therefore, this study 
explores the use of data mining on higher education unit record data to develop a decision 
tree classification model of student success.   
 
 
 
 
 
ii 
 
Acknowledgements 
I would like to acknowledge the invaluable instruction, inspiration and guidance 
provided by Dr. Alina Lazar, Youngstown State University, in the field of data mining 
and the development and refinement of this project. 
In addition, I would like to thank Dr. John Sullins for taking over mid-project, the 
responsibilities of thesis advisor and Mr. Thomas Bodnovich and Mr. Robert Hogue for 
their guidance, support, and inspiration and for serving on my thesis committee.    
Finally, I wish to express my sincere thanks to the University of Waikato, 
Hamilton, New Zealand for making the Weka data mining software available online and 
free to the public; and to Ian H. Witten & Eibe Frank for their work developing 
instructions on using the Weka tool. 
 
 
 
 
 
 
 
 
 
 
 
Using Data Mining to Model Student Success 
 
1 
 
1.   Motivation and Goals 
1.1  Motivation 
 While politicians in both the state and federal governments are in agreement that a 
college education is necessary for economic recovery and technological superiority in the 
world, parents feel that a college education is necessary to obtain worthwhile 
employment.  In Ohio, higher education institutions are continually hearing that business 
and industry are looking for an educated workforce.  The belief is that if the state 
educates and retains those graduates, the jobs will come and the economy will flourish.
 Moreover, given the current state of our nation’s economy, higher education is 
being asked to justify the expense of a college degree.  Parents are concerned that it is 
taking longer than the anticipated four years for their son/daughter to earn a four-year 
degree.  The rising cost of tuition puts an added strain on household budgets and has 
many draining their savings.  Skyrocketing student loan debt has student’s pondering the 
continual dilemma of going further into debt, getting a job or working more hours and 
reducing their academic load, or dropping out of school all together.   
In July of 2009, the President’s Council of Economic Advisers (CEA) published a 
report forecasting employment opportunities for the next decade and outlining the 
groundwork required to prepare the labor force for the new millennium.  The report 
posits that “well-trained and highly-skilled workers will be best positioned to secure 
high-wage jobs” and that “occupations requiring higher educational attainment are 
projected to grow much faster than those with lower education requirements.”  In support 
of that prediction the report goes on to state that the most current job growth has been 
experienced in those professions demanding higher education credentials and job loss in 
Using Data Mining to Model Student Success 
 
2 
 
professions with lesser demands (Executive Office of the President, Council of Economic 
Advisors, 2009). 
Given the downturn in the U.S. economy and subsequent 10%+ unemployment 
rate currently being experienced throughout the nation, this is the time when higher 
education can have its greatest positive impact.  It is common knowledge that during 
times of economic recession, higher education experiences enrollment growth.  In 
addition to current high school graduates pursuing their academic dreams for a better 
future, the recently unemployed or under-employed seek opportunities to acquire or 
refine skills necessary to be competitive in the diminished job market.  “Therefore we 
need a comprehensive strategy to ensure that our education and training systems are 
strong and effective” (Executive Office of the President, Council of Economic Advisors, 
2009). 
In order to guarantee that students have worthwhile educational opportunities 
available, the federal and state governments are making strides in funding reformation.  
In particular the state of Ohio is preparing to rollout sweeping changes to how it allocates 
state subsidization of higher education.  “Instead of funding institutions based on the 
number of students they enroll, the new formula would appropriate dollars based on 
colleges’ ability to retain and graduate students” (Moltz, 2009).   This is a step, like many 
made in the past decade, toward greater accountability for colleges and universities and 
may present major challenges for institutions with open-enrollment policies.  But as Dr. 
Watson Scott Swail, President and CEO of the Educational Policy Institute, opined in his 
August 2008 article The Bell Curve Under a Different Cover, “if our system is such that 
Using Data Mining to Model Student Success 
 
3 
 
we let in a broad cross-section of students, then we have a moral and legal obligation to 
do what we can to help those students succeed.”   
Not only are the federal and state governments seeking to hold higher education 
more accountable, students and their parents are as well.  People take notice when stories 
circulate about college graduates landing barely-above-minimum-wage jobs that not only 
do not provide a living wage but also do not provide enough income for graduates to 
meet the scheduled payments of their student loans (Perry, 2008).   It was reported in 
October 2008 that “the latest generation of adults in the United States may be the first 
since World War II, …not to attain higher levels of education than the previous 
generations.”  The biggest declines are being experienced among the minority 
populations where it is believed that “the current generation is, on average, heading 
toward being less educated than its predecessor” (Jaschik, 2008).  Perhaps conscious 
choices are being made to forego higher education because it is no longer perceived to 
have a good return on the investment.  After all, the time required to earn a four-year 
degree seems to extend each year – the current average is near 5.5 years.  Additionally, 
tuition rates for the most part rise every year.  Combine the increased time-to-degree with 
predatory lending practices which target college students with everything from high 
interest educational loans to even higher interest credit cards, and it is no wonder that 
students are often split between spending more time in college and exiting further in debt 
or opting out of college altogether.   
During the past few years across the nation, we have all as a society experienced 
the rapid ascent of utility costs, the push toward becoming “greener,” and the increased 
costs associated with medical coverage.  Higher education has not been immune.  In fact 
Using Data Mining to Model Student Success 
 
4 
 
while tackling those challenges, higher education has also been expected to provide 
“highly-skilled – and often expensive talent,… top-notch academic support, counseling, 
health services, and campus security,… a nicely maintained campus,… up-to-date 
libraries, labs, and other scientific resources” while satisfying government mandated 
responsibilities (Jacobs and Hyman, 2009).  Higher education also finds itself confronted 
by a public that is crying out for if not a tuition freeze then a reduction in tuition costs.   
1.2 Goals 
In an effort to identify a solution to the problems presented, I chose to employ 
data mining - specifically the construction of a decision tree.  Different from traditional 
statistics, which calculate the probability of a specific hypothesis, the end result of data 
mining is to identify the hidden pattern of connections within the data.  Decision trees are 
just one several techniques used to display those patterns.  Beginning with the attribute 
found by the data mining software to be the most significant and branching out from 
there, a decision tree provides a visual tree-like representation of the underlying 
connections leading to the final outcome.  
By developing a decision tree model of student success, defined here as earning a 
baccalaureate/bachelor degree within six years, this thesis attempts to provide 
information necessary for determining how to increase the graduation rate at a public, 
four-year, open-enrollment institution.  Additionally this increase would benefit the 
institution by satisfying the calls for accountability, optimizing institution’s share of state 
and federal financial assistance, and hopefully make timely degree completion the norm. 
 In order to address the issue of how to increase the percentage of students who 
complete their studies and earn a baccalaureate (four-year) degree within six years, I 
Using Data Mining to Model Student Success 
 
5 
 
attempted to identify the qualities of those students who earn their degrees in a timely 
manner.  Next I used those qualities or criteria to examine a cohort of incoming students 
to predict which students should be successful and followed their progress.  For those 
students who graduated, this information needs to be shared as much as possible with 
high school guidance counselors, so that high schools will know what preparation is 
necessary for their students to succeed in higher education.  In order to best leverage 
limited recruiting dollars and guarantee critical completion-based state subsidy, focus 
should be placed on prospective students that exhibit these identified qualities.  In 
addition, if students were predicted to graduate but did not, then these salient qualities or 
factors should also be identified.  Examining student data may illuminate the key 
indicators that provide vital information on when, and in what manner, intervention 
should occur to facilitate the student’s goal of obtaining a four-year degree.  Students 
who are predicted not to graduate need to be further studied to identify how best to help 
them achieve success – perhaps fueling an argument for selective admissions or increased 
funding of student services.  
 To examine the multitude of student data and identify the relationships between 
the attributes, a decision tree will be constructed using the open-source data mining 
software package - Weka.  It is the intent of this study to utilize the robust computing 
power of data mining software to quickly and accurately predict student success. 
 The organization of this paper follows:   
1. Identify where applicable data are collected and at what frequencies.  
2. Explain why a particular subset of data was chosen. 
3. Tell how that subset was augmented to create the student cohort dataset.  
Using Data Mining to Model Student Success 
 
6 
 
4. Detail how and where the data were retrieved. 
5. Disclose how the data were processed for analysis.   
6. Display the frequency percentage distributions of the student cohort attributes. 
7. Provide a synopsis of the data mining package selected. 
8. Show the results of the application of the decision tree algorithm. 
9. Breakdown the comprehensive analyses of the predicted versus actual 
outcomes by attribute.    
10. Discuss the accuracy of the predictions and possible explanation for the 
unexpected results in some sub-categories in comparison to the research of 
others.  
11. Share conclusions on why this method may be used to help meet the 
challenges facing higher education today. 
12. Proffer recommendations for further research and analysis. 
2.  Data Collection and Processing 
2.1  Data Colletion 
In 1998, the Ohio Board of Regents, the state’s higher education governing body, 
replaced its antiquated higher education data collection process, the Uniform Information 
System, with an updated and expanded data collection system, the Higher Education 
Information System (HEI).  “The Higher Education Information (HEI) system contains 
data supplied by Ohio's colleges and universities.  It is a comprehensive relational 
database that includes data on students, courses, faculty, facilities, and finances” (Ohio 
Board of Regents, 2009).  The HEI System consists of several data modules, or primary 
data areas:  Academic Programs; Enrollment; Facilities; Faculty-Staff; Financial; State 
Using Data Mining to Model Student Success 
 
7 
 
Grants and Scholarships (SGS) Financial Aid; and recently implemented Unit Record 
Tuition and Financial Aid.   
Of interest in this study are the data submitted through the Enrollment data area.  
Data in this region include student demographic data (e.g., birth date, gender, 
race/ethnicity), state and county of residency, student enrollment data (e.g., course title, 
catalog number, credit hour for every course a student is enrolled in a given academic 
term), and student degree/certificate information.  While some data elements are 
collected on an academic term basis (student and course enrollments) others are collected 
annually (degrees/certificates awarded).  In addition to collecting a plethora of data 
values, the HEI System provides web access to an institution’s submitted data to 
authorized users.   This large volume of data can be used for verification purposes, 
benchmarking, longitudinal/trend analyses and in this case, data mining.   
As Bailey (2006) stated, “with any kind of databases that contain 
multidimensional subjects and span multiple years, data mining is an ideal approach to 
identify hidden patterns and discover future trends of behaviors.”   Therefore, the HEI 
System serves as an excellent repository for and source of valuable and extensive data 
critical for this purpose.  Because developing an accurate student success model using 
data mining requires a set of consistently captured data accumulated during the tracking 
of a specific entering cohort of students over a six-year period of time, a compilation of 
data related to first-time undergraduate students entering a Northeastern Ohio, public, 
four-year institution in 2001 was utilized in this research.  Using the menu of predictors 
identified in the decision tree-related research by Herzog (2006), a subset of readily 
available HEI data elements was compiled.  This subset was later supplemented with data 
Using Data Mining to Model Student Success 
 
8 
 
from the institution’s legacy data system in order to address college-readiness, student 
support (tutoring, supplemental instruction, etc.), and student financial need and financial 
aid awards. 
2.2  Data Processing 
The data files were initially downloaded from the HEI website or were obtained 
by querying the institution’s legacy database.  A university-issued student identifier 
served as a primary key between the files and facilitated the population of a cohort table 
stored in MS Excel spreadsheet format for processing.  Each field was reviewed for its 
possible contribution to the classification model.  If it became apparent that some fields 
were redundant, those fields were subsequently removed from the cohort table.  In at least 
one case the range of different numerical data values for a specific field reached over 
100.   Because data mining algorithms typically examine each value looking for patterns 
within the data, fields with a large range of possibilities (e.g., federal financial aid 
awards, state financial aid awards, federal Work Study aid awards, student majors) were 
pared down through a discretization process to reduce the amount of differentiations and 
increase the accuracy of the classification model.   Once the final table was populated, the 
table was converted to a comma separated value file required by Weka, the data mining 
package used in this study. 
2.3  The Cohort 
 
 Cohort attributes were divided into five subcategories: General demographics; 
College readiness; Socio-economic/Financial data; Academic Ability; and Retention. 
Using Data Mining to Model Student Success 
 
9 
 
• General demographics included: gender; racial background; age; state 
residency; and housing status as of the first term of enrollment.  
• College readiness included: ACT composite scores; high school 
graduating grade point averages; advanced placement credits; and 
academic intention (first term major could also be included to the extent 
that it is an indicator of an incoming student’s academic focus).   
• Socio-economic/Financial data included: student marital status (which one 
could argue is also a general demographic); financial dependence upon 
parents; calculated cost of attendance; 9-month estimated expected family 
contribution; financial aid determined need level; federal financial aid 
(excluding student loans); state aid; federal Work Study aid; institutional 
aid; other third party aid; and student loans.  
• Academic ability was comprised of: first term academic load; attempted 
credit hours; credit hours earned; total quality points; and end of term 
grade point average; engagement in remedial English; remedial 
mathematics or Reading & Study Skills developmental coursework; and 
number of visits paid to the Center for Students Progress for assistance 
with peer mentoring, tutoring, and the like. 
• Retention, included: returning the following spring term; continuous 
enrollment from entry fall term, to subsequent fall term including spring 
(but not accounting for summer term); and returning the following fall 
term. 
The distributions of the values for these data are shown on the following charts. 
Using Data Mining to Model Student Success 
 
10 
 
 
Figure 1 
          As has been the trend at this institution, just slightly more females than males (53% 
vs. 47%) were found in the 2001 cohort.   
 
Figure 2 
          The clustering of years of birth around 1982 and 1983 indicates that most of the 
students in the cohort were in the traditional age group for college/university students.  
Nearly 86% of the students in the cohort were 19 years of age or younger.   
52.92%
47.08%
0%
10%
20%
30%
40%
50%
60%
70%
Total
Gender
Female
Male
85.99%
14.01%
0%
20%
40%
60%
80%
100%
Total
Age Groups
19 or Younger
20 or Older
Using Data Mining to Model Student Success 
 
11 
 
 
 
Figure 3 
          Slightly more than 80% of the students in the cohort were White.  The next 
frequency in the category was Black with slightly less than 10%. 
 
Figure 4 
          Close to 90% of the students were state residents at the time of application. 
 
0.45% 0.54%
9.55%
2.13%
0.59%
6.29%
80.45%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Total
Racial/Ethnic Background
American_Indian
Asian
Black
Hispanic
International
Unspecified_Race
White
89.65%
10.35%
0%
20%
40%
60%
80%
100%
Total
State Residency
Ohio Resident
Not an Ohio Resident
Using Data Mining to Model Student Success 
 
12 
 
 
 
 
 
 
 
 
 
 
           
A minority of 18.6% of the cohort population fell into the campus resident 
category for the first term of enrollment. 
 
Figure 6 
          A majority of the cohort (over 80%) submitted an ACT composite score.  Of that 
group 71% had a composite score of 18 or higher. 
0.30%
23.12%
40.35%
15.64%
1.53%
19.06%
0%
10%
20%
30%
40%
50%
60%
Total
ACT Composite Score Ranges
6-11
12-17
18-23
24-29
30-36
No ACT
18.61%
81.39%
0%
20%
40%
60%
80%
100%
Total
Housing Status
Campus Resident
Commuter
Figure 5 
Using Data Mining to Model Student Success 
 
13 
 
 
Figure 7 
          Most students entering in fall of 2001, submitted high school graduating overall 
grade point averages of below a 3.0.  
 
Figure 8 
          Nearly the entire student cohort group did not have any advance placement credits 
at the time of enrollment. 
57.16%
40.95%
1.88%
0%
20%
40%
60%
80%
100%
Total
High School Graduating GPAs
Below 3.0 HS GPA
3.0 or Above HS GPA
GED_recipient
97.97%
2.03%
0%
20%
40%
60%
80%
100%
Total
Advanced Placement Credits
No Advanced Placement 
Credit
Advanced Placement 
Credit
Using Data Mining to Model Student Success 
 
14 
 
 
Figure 9 
          At the time of initial application to the institution, the majority of students in the 
cohort indicated that they intended to either obtain a bachelor’s degree or that they did 
not know what their academic intention was (42.4% and 47% respectively).   
2.28%
0.10%
42.38%
0.10%
5.50%
1.29%
0.69% 0.69%
46.98%
0%
10%
20%
30%
40%
50%
60%
Total
Academic Intention
Obtain_Associate_Degree_for_Job_Market Obtain_Associate_Degree_for_Transfer
Obtain_Bachelors_Degree Obtain_Undergraduate_Certificate
Personal_Interest Selected_Courses_Train_New_Career
Selected_Courses_Upgrade_Skills Transfer_Before_Degree
Unknown
Using Data Mining to Model Student Success 
 
15 
 
 
Figure 10 
          Though first term fields of study were widely dispersed, the Professional & Applied 
Sciences group held the most popular majors. 
 
 
 
 
 
 
 
6.09%
11.19% 10.94%
5.50%
61.39%
4.90%
0%
10%
20%
30%
40%
50%
60%
70%
Formal 
Sciences
Humanities Liberal Arts Natural 
Sciences
Professional & 
Applied 
Sciences
Social Sciences
Major Field of Study
First Term
Formal Sciences Humanities
Liberal Arts Natural Sciences
Professional & Applied Sciences Social Sciences
0.35%
2.18%
97.48%
0%
20%
40%
60%
80%
100%
Total
Marital Status
Life_Partner
Married
Single
Figure 11 
Using Data Mining to Model Student Success 
 
16 
 
Of the students in this study (see Figure 11), approximately 97% were single, 
slightly more than 2% were married and less than 1% reported living with a life partner. 
 
Figure 12 
          More than 2/3 of the cohort population was financial dependent upon their parents 
when they first entered the university. 
68.66%
9.01%
22.33%
0%
20%
40%
60%
80%
100%
Total
Financially Dependent Upon 
Parents
Dependent
Independent
Unknown
Using Data Mining to Model Student Success 
 
17 
 
 
Figure 13 
          For those students submitting a valid FAFSA (Free Application for Federal Student 
Aid), the majority (49%) had an annual estimated cost of attendance equal to that of a 
full-time, in-state student. 
49.01%
4.95%
37.52%
5.88%
2.64%
0%
10%
20%
30%
40%
50%
60%
Total
Cost of Attendance
(For those with a Valid FAFSA)
2. 9001-10000
3. 10000-12000
4. 12001-14000
5. 14001-16000
6. 16001-18000
Using Data Mining to Model Student Success 
 
18 
 
 
Figure 14 
          The 9-Month Expected Family Contribution (EFC) is the value the federal 
government believes a student’s family is capable of paying as a result of calculations 
based upon the student’s responses on the FAFSA.  Most students had a 9-month 
expected family contribution of $3,000 or greater. 
30.15%
69.85%
0%
20%
40%
60%
80%
Total
9-Month Estimated Expected 
Family Contribution
(For those with a Valid FAFSA)
Less than $3,001
$3,001 or Greater
Using Data Mining to Model Student Success 
 
19 
 
 
Figure 15 
          Need Level values are determined by examining the Cost of Attendance with 
respect to the 9-Month Estimated Family Contribution.  The range of student need levels 
spanned from –$77,999 to over $14,001 with a majority of the students (over 80%) 
showing a financial need over$1 and up to over $14,001.  It is noteworthy to point out 
that over 46% of the students showed a need level of over $8,000. 
1.72%
3.04%
14.93%
4.76%
12.35%
17.04%
21.66%
6.87%
15.46%
2.18%
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
Need Level Ranges
(For those with a valid FAFSA)
Using Data Mining to Model Student Success 
 
20 
 
 
Figure 16 
          Nearly all students received some amount of financial aid. 
 
Figure 17 
          More than half of the cohort did not receive any federal financial aid.  Of those 
receiving federal financial aid, most received $1,000 or more. 
99.34%
0.66%
0%
20%
40%
60%
80%
100%
Total
Received Any Financial Aid
Any Aid - Yes
Any Aid - No
52.31%
13.41%
34.28%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Total
Federal Financial Aid
(For those with a Valid FAFSA)
1. No_Aid
Between $1 and $999
$1000 and Over
Using Data Mining to Model Student Success 
 
21 
 
 
Figure 18 
          Just about 40% of the student population in this study received state aid.  Most of 
those receiving state aid were awarded $500 or less. 
 
Figure 19 
          Less than 2% of the cohort participated in the federal work study program, which 
sponsors part-time jobs for students to earn money for college.  
59.97%
20.01%
10.77%
8.72%
0.46%
0.07%
0%
10%
20%
30%
40%
50%
60%
70%
Total
State Aid
(For those with a Valid FAFSA)
1. No_State_Aid
2. 1-500
3. 501-1000
4. 1001-1500
6. 2001-2500
7. Over 2500
98.12%
1.88%
0%
20%
40%
60%
80%
100%
Total
Federal Work Study Aid
(For those with a Valid FAFSA)
No Work Study Aid
Awarded Work Study 
Aid
Using Data Mining to Model Student Success 
 
22 
 
 
Figure 20 
          Institutional Aid includes fee remission for current and retired employees, their 
spouses and dependents, and internal and external scholarships administered by the 
institution.  Just about 40% of the cohort student group received some amount of 
institutional aid. 
59.31%
21.86%
18.82%
0%
20%
40%
60%
80%
100%
Total
Those Receiving Institutional Aid
No_Institutional_Aid
Between $1 - $500
Over $500
Using Data Mining to Model Student Success 
 
23 
 
 
Figure 21 
          Agency funded money for retraining and local independent awards categorized as 
Other Third Party Aid was received by almost all (94%) of the students in the cohort. 
 
Figure 22 
6.21%
79.66%
14.13%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Total
Other Third Party Aid
No Third Party Aid
Between $1 and $600
$601 or More
42.87%
33.95%
16.78%
5.15%
1.12%
0.13%
0%
10%
20%
30%
40%
50%
60%
70%
Total
Student Loan Ranges
No Student Loan
$2,000 or Less
$2,001 to $4,000
$4,001 to $6,000
$6,001 to $8,000
$8,001 to $8,500
Using Data Mining to Model Student Success 
 
24 
 
          More than half (57%) of the cohort population took out a student loan during their 
1
st
 academic year of study – most in the amount of $2,000 or less. 
 
Figure 23 
          Almost all (94%) of this special population engaged in a full-time academic load of 
12 or more credit hours of course work. 
 
Figure 24 
94.06%
5.94%
0%
20%
40%
60%
80%
100%
Total
First Term Academic Load
Full-Time
Part-Time
4.50% 4.16%
17.18%
61.19%
12.97%
0%
20%
40%
60%
80%
100%
Total
First Term Attempted Credit 
Hours 
1. 0
2. 1-5
3. 6-10
4. 11-15
5. 16-19
Using Data Mining to Model Student Success 
 
25 
 
          The distribution of first term attempted credit hours indicates that most students 
were enrolled between 11 – 15 credit hours in for credit courses.  As this study attempts 
to predict students entering at a specified point and earning a bachelor’s degree within six 
years, it is important to point out that in order to earn the required average of 126 credit 
hours for a bachelor degree within the specified timeframe, students need to complete 
roughly 16 credit hours per term to graduate within four academic years, 13 credit hours 
per term to graduate within five years, and 11 credit hours per term to graduate within six 
years.  Additionally, in order to receive the maximum amount of financial aid a student is 
eligible to be awarded, a student must be enrolled in at least 12 credit hours of course 
work.   
          Note that 4.50% of the cohort population was engaged in coursework, in all 
likelihood auditing courses, which upon completion would earn them no academic credit.  
Further examination of this group may reveal that the academic intention of this group of 
students indicated that they were seeking something other than an academic certificate or 
degree. 
Using Data Mining to Model Student Success 
 
26 
 
 
Figure 25 
          Where only 4.5% of the cohort knowingly enrolled in course work applicable for no 
academic credit, nearly 8% more (or 12.7%) actually earned no academic credit.  This  
additional 8% can be attributed to students completely withdrawing from all their 
coursework after the enrollment census point, students failing all of their coursework for 
the term, or students who failed to officially withdraw from the institution earning non-
attendance failing grades. 
12.67%
26.19%
52.87%
8.27%
0%
10%
20%
30%
40%
50%
60%
Total
Credit Hours Earned
No Hours
1.00 to 11.00
12.00-16.00
17.00+
Using Data Mining to Model Student Success 
 
27 
 
 
Figure 26 
          During the first term of enrollment for this cohort the institution had no policy in 
place for distinguishing those students earning non-attendance failing grades from those 
students actually earning failing grades. Embracing the belief that at least half of the 
students in the 0 quality points range either officially withdrew after the census point or 
earned non-attendance failing grades, the quality points (the values 4 through 0 assigned 
in accordance with letter grades earned in course work) earned resemble the normal 
distribution.   
13.51%
9.65%
15.00%
19.11%
23.32%
15.89%
3.51%
0%
10%
20%
30%
40%
50%
60%
Total
First Term Total Quality Points 
Earned
1. 0
2. 1-12
3. 13-24
4. 25-36
5. 37-48
6. 49-60
7. 61-76
Using Data Mining to Model Student Success 
 
28 
 
 
Figure 27 
          At the conclusion of the first academic term of study, more than half of the students 
in the cohort had grade point averages below 3.0 (61% vs. 39%). 
 
Figure 28 
          Interestingly, irrespective of placement testing recommendations, more than half 
(55%) of this population at sometime during their academic careers engaged in remedial 
coursework.   
17.62%
13.51%
14.41%
15.35%
11.49%
9.16%
7.23%
11.24%
0%
5%
10%
15%
20%
25%
30%
35%
Total
First Term GPAs
1. Below 1.0
2. 1.00-1.99
3. 2.00-2.49
4. 2.50-2.99
5. 3.00-3.24
6. 3.25-3.49
7. 3.50-3.74
8. 3.75 and higher
44.80%
55.20%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Total
Any Remediation
No
Yes
Using Data Mining to Model Student Success 
 
29 
 
 
Figure 29 
          Just about 37% of the population engaged in remedial English. 
 
Figure 30 
          Likewise nearly 37% of the population engaged in remedial mathematics. 
 
 
 
 
63.12%
5.10%
31.78%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Total
Remedial English
No Remedial English
Failed Remedial English
Passed Remedial 
English
62.92%
10.79%
26.29%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Total
Remedial Mathematics
No Remedial Math
Failed Remedial Math
Passed Remedial Math
Using Data Mining to Model Student Success 
 
30 
 
 
 
 
 
 
 
 
 
 
  
 
          And a little more than 21% of the population engaged in Reading & Study Skills 
course work specifically “designed to develop students’ skills essential for college 
studying” and believed to assist underprepared students in achieving a state of college 
readiness (YSU’s Undergraduate Catalog, 2009). 
          Of the three categories of remedial coursework, remedial mathematics had the 
largest amount of students engaged with 749 out of the 2,020 in the cohort.  This figure is 
just four more students than the number engaged in remedial English. 
78.61%
2.52%
18.86%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Total
Reading & Study Skills
No Reading & Study 
Skills
Failed Reading & Study 
Skills
Passed Reading & Study 
Skills
Figure 31 
Using Data Mining to Model Student Success 
 
31 
 
 
Figure 32 
          Only 15% of the cohort population visited the Center for Student Progress during 
their first term, thus taking advantage of any of the numerous the services they provide 
(e.g. tutorial, individual intervention, or supplemental instruction services) for helping 
students “acquire the skills and knowledge needed to become successful learners” 
(Center for Student Progress, 2009). 
 
Figure 33 
15.25%
84.75%
0%
20%
40%
60%
80%
100%
Total
Center for Student Progress
Visited the Center for 
Student Progress
Never Visited the Center 
for Student Progress
84.90%
15.10%
0%
20%
40%
60%
80%
100%
Total
Returned Spring Term
Spring - Yes
Spring - No
Using Data Mining to Model Student Success 
 
32 
 
          A majority (85%) of the entering student population continued to be enrolled the 
following spring term. 
 
Figure 34 
          And slightly more than 67% of the entering cohort was enrolled the first three 
consecutive terms (fall 2001, spring 2002 and fall 2002).   
 
Figure 35 
67.28%
32.72%
0%
20%
40%
60%
80%
100%
Total
Returned Spring and Next Fall
SP & AU - Yes
SP & AU - No
68.76%
31.24%
0%
20%
40%
60%
80%
100%
Total
Returned the Next Fall
Next Fall - Yes
Next Fall - No
Using Data Mining to Model Student Success 
 
33 
 
          Additionally almost 69% returned to the institution the following fall term.  This 
figure is down roughly 16% from those who continued through the first spring but up 
slightly from those students enrolled consecutively fall, spring and fall of the subsequent 
year – indicating that some students not enrolled the immediately following spring term 
do in fact still return the following fall.  
2.4  Data Mining Tool – Weka 
 
Herzog (2006) found that “when working with large data sets to estimate 
outcomes with many predictor variables, data-mining methods often yield greater 
prediction accuracy, classification accuracy, or both [than that of traditional statistics]”.  
Therefore rather than perform typical statistical analyses, data mining, in particular free-
to-the-public, open source data mining software, was employed for this endeavor.   
The open source data mining software, Weka, which stands for Waikato 
Environment for Knowledge Analysis, is a machine learning project undertaken by The 
University of Waikato.  The primary goal of the “project is to build a state-of-the-art 
facility for developing machine learning (ML) techniques and to apply them to real-world 
data mining problems.”  The software is actually a “workbench” of commonly known 
algorithms accessible through four different interfaces (The University of Waikato, 
Using Data Mining to Model Student Success 
 
34 
 
2009).  The four interfaces are accessible via the Weka GUI Chooser.
 
Figure 36 
The Simple command line interface, or Simple CLI, is a text based interface to the 
workbench which requires the user to already be familiar with the software’s facilities. 
The second option, the Knowledge Flow interface, requires an extensive amount of main 
memory to operate is consequently useful in analysis of small- to medium-sized datasets.  
It allows you to drag and drop icons representing the different algorithms on to the screen 
and design your own custom configurations for streamed data processing, again requiring 
some strong working knowledge of data analysis.  The Experimenter interface provides 
assistance in determining which parameter values and algorithms will produce the 
strongest result for the problem at hand.  And finally the Explorer interface, allows a 
Using Data Mining to Model Student Success 
 
35 
 
novice user to easily upload a dataset and employ any of the software’s features via menu 
selections and dropdown lists.   
The software has been developed with an easy-to-use, intuitive style.  Interface 
forms are set up to guide the user through the necessary steps in an appropriate order and, 
like other commercially available software packages, grey-out the selection items that are 
not available under the present conditions (Witten & Frank, 2005).  The Explorer 
interface was used exclusively to perform the analysis on this project. 
 
Figure 37 
 Upon entrance into the Explorer interface, the user must open an appropriately 
formatted data file.  Weka accepts many types of data files, including comma-delimited 
Using Data Mining to Model Student Success 
 
36 
 
(.csv) files.  In this study, .csv files were compiled because of the ease of formatting 
available with MS Excel 2007.  After opening the data file the user is able to view and as 
needed remove attributes from the dataset via the Attributes window.  This feature 
facilitates analysis by removing the attribute only from the Weka interface and not from 
the underlying dataset. 
  
Figure 38 
 Once the desired attribute listing has been compiled, the user seeking to develop a 
decision tree then clicks on the Classify tab at the top of the screen to enter the next phase 
of processing.  Here the user is able to access the many available classifier algorithms by 
clicking on the Choose button (not visible in this screen shot.) 
Using Data Mining to Model Student Success 
 
37 
 
  
  
Figure 39 
 Next the user selects the desired evaluation options.  It is important in this 
window to be sure to check the Output model, Output per-class stats, Output confusion 
matrix, and to Choose the file type for the Output predictions.  These predictions are later 
appended to the original dataset in order to facilitate the development of MS Excel pivot 
tables and subsequent pivot charts for presenting the data analysis.  Then from the Test 
Options dropdown box the user selects the target attribute, in this case Bachelor degree.  
Then the user clicks the Start button.  
Using Data Mining to Model Student Success 
 
38 
 
 
Figure 40 
Within a few minutes, Weka produces the selected output and feeds it back to the 
user screen.   
3.  Analysis 
The initial dataset contained attributes influenced by the published research of 
many in the field of institutional research supported by the recently published work of 
Bowen, Chingos and McPherson (2009).   A majority of the data elements were selected 
based on their availability within the first academic year of study (e.g. ACT Composite 
test score, high school graduating grade point average, first term student credit hour load, 
returned spring term, etc.).  For the data mining output, in this case a decision tree, to 
provide a meaningful predictor of future student success, it is important that the dataset 
Using Data Mining to Model Student Success 
 
39 
 
be comprised of attributes significant to the accurate prediction of outcome as early as 
possible in a student’s academic career - thus, affording the institution time to intervene.   
Because “each technique employs a learning algorithm to identify a model that 
best fits the relationship between the attribute set and class label of the input data” (Tan, 
Steinbach & Kumar, 2006) , after determining the list of data elements desired for 
building the model, Weka was employed to process the data using the available decision 
tree classifiers.  The J48 algorithm produced the strongest accuracy based on the initial 
dataset and was therefore chosen for developing the final decision tree model.  Note that 
in order to increase the precision of the predictions, some original dataset attributes were 
removed or modified and others introduced. 
 
 
Figure 41 
Once the beginning stages of analysis yielded less accurate results than expected, 
a re-evaluation of the dataset attributes took place.    
86.29%
13.71%
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
Correct
Predictions
Incorrect
Predictions
J48's Success Rate
Using Data Mining to Model Student Success 
 
40 
 
The initial dataset consisted of the following data elements: 
o Gender 
o Year of birth 
o Ethnicity/race 
o Zip code 
o Academic intent 
o Student rank 
o State residency 
o First term attempted credit hours 
o First term cumulative quality points 
o First term cumulative grade point average 
o First term cumulative total credit hours 
o Major code 
o Living arrangements 
o Federal financial aid (excluding student loans) 2001-02 
o State financial aid 2001-02 
o Federal Work Study aid 2001-02 
o Student loans 2001-02 
o Institutional aid 2001-02 
o Other third party aid 2001-02 
o Dependency upon parents 2001-02 
o Parent marital status 2001-02 
o Student marital status (FAFSA) 2001-02 
o Student marital status FAFSA code 2001-02 
o Parental family size 2001-02 
o Cost of attendance 2001-02 
o 9-month estimated expected family contribution 
o Need level 
o Student marital status from legacy system 
o Academic load 
o High school CEEB code 
o High school graduation year 
o High school class standing 
o Number of students in high school graduating class 
o Advanced placement credit – Biology 
o Advanced placement credit – Chemistry 
o Advanced placement credit – English 
o Advanced placement credit – Foreign Language 
o Advanced placement credit – History 
o Advanced placement credit – Math/Statistics 
Using Data Mining to Model Student Success 
 
41 
 
o Associate degree earned in 2 years, 3 years, 4 years, 5 years, 6 years, or 7 
years  
o Bachelor degree earned in 2 years, 3 years, 4 years, 5 years, 6 years, or 7 
years 
o Master degree earned in 5 years, 6 years, or 7 years  
o Post Baccalaureate certificate earned in 6 years or 7 years 
o Undergraduate certificate earned in 4 years, 5 years, or 6 years 
o Visited the Center for Student Progress (yes/no), visited 1 - 5 times, 
visited 6 – 10 times, visited 11-15 times, visited 16-20 times, visited 21+ 
times 
o Passed remedial English 1540T 
o Failed remedial English 1540T 
o Passed remedial English 1540 
o Failed remedial English 1540 
o Passed remedial math 1501 
o Failed remedial math 1501 
o Passed Reading & Study Skills 1510B 
o Failed Reading & Study Skills 1510 B 
o Passed Reading & Study Skills 1510A 
o Failed Reading & Study Skills 1510A 
 
After the initial data mining process was employed  
the following data elements were removed: 
• student zip code at time of application  
• student rank, parental marital status  
• student marital status (from the FAFSA form) 
• parental family size 
• high school graduation year 
• high school class standing 
• number of students in high school graduating class 
• earned a master degree in 5 years, 6 years or 7 years 
• earned a post baccalaureate certificate in 6 years or 7 years 
• earned an undergraduate certificate in 4 years, 5 years or 6 years 
 
the following data elements were discretized: 
 
o year of birth – age ranges 
o federal financial aid (excluding student loans) 
o major field of study – first term 
Using Data Mining to Model Student Success 
 
42 
 
o state financial aid 
o federal Work Study aid 
o student loans 
o institutional aid 
o other third party aid 
o cost of attendance 
o 9-month expected family contribution 
o need level 
o advanced placement (AP) credits in biology, chemistry, English, foreign 
languages, history, or mathematics to a dichotomous (yes/no) field for any 
AP credits 
o earned an associate degree in 2 years, 3 years, etc. to a dichotomous 
(yes/no) field for earned an associate degree ever 
o earned a baccalaureate degree in 2 years, 3 years, up to 6 years to a 
dichotomous (yes/no) field for earned a baccalaureate within 6 years 
o visited the Center for Student Progress 
o failed remedial English and passed remedial English to a trichotomous 
field (did not take, failed, passed) 
o failed remedial mathematics and passed remedial mathematics to a 
trichotomous field (did not take, failed, passed) 
o failed Reading & Study Skills and passed Reading & Study Skills to a 
trichotomous field (did not take, failed, passed) 
 
the following data elements were introduced: 
 
• returned the immediately following spring term 
• continued through spring and fall terms 
• returned the subsequent fall term 
• any financial aid* 
• completed the FAFSA* 
• any AP credits* 
• any remediation* 
 
*added during the final analysis stage to provide further context for appropriate 
interpretation 
 
After incorporating changes in the dataset to increase the precision of the 
algorithm, the Weka software using the J48 decision tree classifier was able to achieve an 
86.29% accuracy rate on student success predictions for the 2020 instances in the fall 
Using Data Mining to Model Student Success 
 
43 
 
2001 student cohort.  The decision tree J48 produced utilizing the training data is very 
large - 272 branches with 227 leaves.  As explained in the Introduction to Data Mining 
text book (Tan et al., 2006) by splitting the branches so many times, J48 may be 
overfitting the solution specifically to the training set data and may yield a lesser level of 
accuracy when applied to future datasets.  Typically data mining software is invoked for 
processing extensive amounts of data with a large number of instances.  The guiding 
principle behind data mining is that enormous amounts of data provided for analysis 
afford the data mining algorithm to learn which attributes are meaningless in predicting 
the outcome allowing the algorithm to prune those branches from the tree.   The result is 
a smaller decision tree with greater prediction accuracy.   
Therefore in this case it is believed that the enormity of this decision tree is due to 
the fact that the dataset itself was quite small – only 2020 instances; forcing J48 to split 
the tree into multiple branches in order to classify each instance.  (See Appendix A for 
the results of the application of the J48 algorithm in Weka.)  Methods for increasing the 
accuracy of decisions trees like boosting (Roe, Yang, Zhu, Liu, Stancu, & McGreagor 
2004) or windowing (Long, Griffith, Selker, & D’Agosino, (1993) may provide avenues 
for future research. 
 
 
Using Data Mining to Model Student Success 
 
44 
 
A Simplified Version of the Resulting J48 Decision Tree 
 
Figure 42 
Returned 
Next Fall
# of CSP 
Visits
Yes/No
Federal 
Aid
Yes/No
Cum 
Quality 
Point 
Ranges
Yes/No
Age 
Range
Yes/No
9‐Month 
EFC
Yes/No
Composite ACT 
Score
Yes/No
Cum 
Credit 
Hour 
Range
Yes/No
# of CSP 
Visits
Yes/No Major
Yes/No Ethnicity
Yes/No
HS CEEB 
Code
Yes/No
State 
Resident
Yes/No
Other 3rd 
Party Aid
Yes/No
Federal 
Aid
Yes/No
HS CEEB 
Code
Yes/No
Remedial 
Math
Yes/No Institutional Aid
Yes/No
HS CEEB 
Code
Yes/No
Remedial 
English
Yes/No
Academic 
Intent
Yes/No
Other 3rd 
Party Aid
Yes/No
Gender
Yes/No
HS CEEB 
Code
Yes/No
PID
Yes/No
State Aid 
Range
Yes/No
Federal 
Aid
Yes/No Institutional Aid
Yes/No Dependency
Yes/No
PID
Yes/No
Academic 
Intent
Yes/No Pid
Yes/No
# of CSP 
Visits
Yes/No Ethnicity
Yes/No
Remedial 
English
Yes/No
Cum GPA 
CrHr
Yes/No Commuter
Yes/No
RD&SK 
Course
Yes/No
State Aid
Yes/No
Associate
Yes/No
Using Data Mining to Model Student Success 
 
45 
 
 
Figure 43 
 
Figure 44
J48 predicted that a little more than 31% of the students in the cohort would earn 
a bachelor degree.  In fact just over 33% actually successfully completed their degree 
requirements and earned a bachelor degree within six years of initial entrance. 
 
Figure 45 
 
Figure 46
Of the 31.53% of the cohort predicted to earn a degree 81.16% did.  Of the 
68.47% predicted to not earn a degree 88.65% did not.  There were 1,383 students 
predicted not to earn a degree in comparison to 637 students predicted to earn a degree.  
31.53%
68.47%
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
Total
Predicted Outcome
Predicted
Yes
Predicted
No
33.37%
66.63%
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
Total
Actual Outcome
Earned
Degree
Did Not
Earn
Degree
81.16%
11.35%
18.84%
88.65%
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
Predicted 
-Yes
Predicted 
-No
J48's Predictions with
Actual Outcome
Earned
Degree
Did Not
Earn
Degree
76.71%
8.92%
23.29%
91.08%
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
Earned
Degree
Did Not
Earn
Degree
Actual Outcome with 
J48s Prediction
Predicted
-Yes
Predicted
-No
 
46 
 
More instances in the Predicted – No category provided J48 with enough examples to 
accurately predict over 89% of the final outcomes.  This result in comparison to the 81% 
accuracy of the Predicted – Yes category provides an illustration of how the data mining 
algorithm obtains a higher level of accuracy with a greater number of instances.   
Of the 674 students actually earning a degree, J48 only correctly predicted 
76.71% of the outcomes.  Of the 1,346 students not earning a degree, J48 correctly 
predicted 91.08% of the outcomes.  These results further support the belief that the more 
instances available for analysis the greater the accuracy of the resulting decision tree.  
The chart of Students Earning a Bachelor Degree predicted versus actual outcomes 
follows as well as charts depicting predicted and actual percentages for each attribute. 
 
Figure 47 
 
 
 
 
0%
5%
10%
15%
20%
25%
30%
35%
Students Earning a Bachelor Degree
Predicted Actual
 
47 
 
 
Figure 48 
 
Figure 49 
J48 performed slightly better predicting 
which female students would or would not 
graduate than it did for the male students. 
 
 
Figure 50 
 
Figure 51 
For both sub-categories, J48’s predictions 
were off by about 2 percentage points. 
 
 
 
 
 
35.73%
26.81%
64.27%
73.19%
0%
20%
40%
60%
80%
Female Male
Gender
Predicted - Yes
Predicted - No
36.95%
29.34%
63.05%
70.66%
0%
20%
40%
60%
80%
Female Male
Gender
Earned Degree
Did not Earn 
Degree
35.22%
9.12%
64.78%
90.88%
0%
20%
40%
60%
80%
100%
19 or 
Younger
20 or Older
Age Groups
Predicted - Yes
Predicted - No
37.00%
11.23%
63.00%
88.77%
0%
20%
40%
60%
80%
100%
19 or Younger 20 or Older
Age Groups
Earned Degree
Did not Earn 
Degree
 
48 
 
 
Figure 52 
 
Figure 53 
For American Indian, J48’s prediction was 
exactly correct.  For Black, International, 
and White, J48 under predicted the 
percentages earning degrees.  For the 
remaining backgrounds, J48 over predicted 
the outcomes – most notably Asian, whose 
outcome was the exact opposite of the 
prediction.   
 
Figure 54 
 
Figure 55 
Just about 3% more state residents and 3% 
less non-state residents earned a bachelor 
degree within six years of initial entrance. 
 
 
 
22.22%
54.55%
9.33%
18.60%
75.00%
26.77%
34.46%
77.78%
45.45%
90.67%
81.40%
25.00%
73.23%
65.54%
0%
20%
40%
60%
80%
100%
Racial/Ethnic Backgrounds
Predicted - Yes
Predicted - No
22.22%
45.45%
17.62%
16.28%
58.33%
31.50%
35.63%
77.78%
54.55%
82.38%
83.72%
41.67%
68.50%
64.37%
0%
20%
40%
60%
80%
100%
Racial/Ethnic Backgrounds
Earned Degree
Did not Earn 
Degree
30.92%
36.84%
69.08%
63.16%
0%
20%
40%
60%
80%
100%
State Resident Not a State 
Resident
State Residency
Predicted - Yes
Predicted - No
33.35% 33.49%
66.65% 66.51%
0%
20%
40%
60%
80%
100%
State Resident Not a State 
Resident
State Residency
Earned Degree
Did not Earn 
Degree
 
49 
 
 
Figure 56 
 
Figure 57 
For commuters the J48 prediction was 
accurate within one percentage point.  For 
campus residents it under predicted those 
earning a degree by approximately two 
percentage points. 
 
 
Figure 58 
 
Figure 59 
More students actually earned a degree than 
were predicted for scores ranging from 6 to 
23.  The unexpected lower graduation rate of 
those with a score of 24 or higher may due 
to students transferring out to another 
institution in pursuit of a program not 
offered at this institution.  Unfortunately that 
information was not available at the time of 
this study. 
39.10%
29.81%
60.90%
70.19%
0%
20%
40%
60%
80%
100%
Commuter Campus 
Resident
Housing Status
Predicted - Yes
Predicted - No
39.89%
31.87%
60.11%
68.13%
0%
20%
40%
60%
80%
100%
Commuter Campus 
Resident
Housing Status
Earned Degree
Did not Earn 
Degree
15.63%
33.87%
59.49%
80.65%
19.48%
100%
84.37%
66.13%
40.51%
19.35%
80.52%
0%
20%
40%
60%
80%
100%
ACT Composite Score Ranges
Predicted - Yes
Predicted - No
16.67%
21.84%
37.30%
54.75%
77.42%
18.18%
83.33%
78.16%
62.70%
45.25%
22.58%
81.82%
0%
20%
40%
60%
80%
100%
ACT Composite Score Ranges
Earned Degree
Did not Earn 
Degree
 
50 
 
 
Figure 60 
 
Figure 61 
Slightly more than 5% of students with HS 
GPAs below 3.0 (≈ 60% of the cohort) 
earned a degree than were predicted.  Where 
about 3% less of those at 3.0 or above 
earned a degree.  This is the first sub-
category identified where the absence of 
transfer out information surfaces as a 
potential critical factor for predictions. 
 
Figure 62 
 
Figure 63 
The predictions for those with advanced 
placement credits were very off - as 12% 
less of the students with advanced placement 
credits (≈2% of the cohort) earned a degree 
than what were predicted.  Those with no 
advanced placement credits were within 
about 2% of the predicted percentage. 
 
16.55%
53.75%
83.45%
46.25%
100%
0%
20%
40%
60%
80%
100%
Below 
3.0 HS 
GPA
3.0 or 
Above 
HS GPA
GED 
Recipient
High School Graduating GPAs
Predicted - Yes
Predicted - No
21.70%
50.85%
4.00%
78.30%
49.15%
96.00%
0%
20%
40%
60%
80%
100%
Below 3.0 
HS GPA
3.0 or 
Above 
HS GPA
GED 
Recipient
High School Graduating GPAs
Earned Degree
Did not Earn 
Degree
30.52%
80.49%
69.48%
19.51%
0%
20%
40%
60%
80%
100%
No Advanced 
Placement 
Credit
Advanced 
Placement 
Credit
Advanced Placement Credits
Predicted - Yes
Predicted - No
32.64%
68.29%67.36%
31.71%
0%
20%
40%
60%
80%
100%
No Advanced 
Placement 
Credit
Advanced 
Placement 
Credit
Advanced Placement Credits
Earned Degree
Did not Earn 
Degree
 
51 
 
 
Figure 64
 
Figure 65
For the two largest subcategories (Unknown – n = 949 and Obtain_Bachelors_Degree – n = 856) 
J48’s predictions were off by 2% and 3% respectively.    The algorithm performed quite well for 
the subcategories of Obtain_Associate_Degree_for_Transfer, Obtain_Undergraduate_Certificate, 
Personal_Interest, and Transfer_Before_Degree. 
26.09%
50.00%
33.41%
50.00%
21.62%
19.23%
21.43%
14.29%
31.93%
73.91%
50.00%
66.59%
50.00%
78.38%
80.77%
78.57%
85.71%
68.07%
0%
20%
40%
60%
80%
100%
Academic Intention
Predicted - Yes
Predicted - No
23.91%
50.00%
36.57%
50.00%
21.62%
23.08%
28.57%
14.29%
32.88%
76.09%
50.00%
63.43%
50.00%
78.38%
76.92%
71.43%
85.71%
67.12%
0%
20%
40%
60%
80%
100%
Academic Intention
Earned Degree
Did not Earn Degree
 
52 
 
 
Figure 66 
 
Figure 67 
The original amount of student majors were downsized to the 6 major groups listed here in 
Figure 67 for ease in visual interpretation.  J48s prediction for the largest sub-category, 
Professional & Applied Sciences, is lower by slightly less than 4 percentage points. 
21.14%
34.96%
29.41%
34.23%
31.69%
34.34%
78.86%
65.04%
70.59%
65.77%
68.31%
65.66%
0%
20%
40%
60%
80%
100%
Formal 
Sciences
Humanities Liberal Arts Natural 
Sciences
Professional 
& Applied 
Sciences
Social 
Sciences
Major Field of Study
First Term
Predicted - Yes
Predicted - No
24.39%
32.30%
30.32%
31.53%
35.40%
30.30%
75.61%
67.70%
69.68%
68.47%
64.60%
69.70%
0%
20%
40%
60%
80%
100%
Formal 
Sciences
Humanities Liberal Arts Natural 
Sciences
Professional 
& Applied 
Sciences
Social 
Sciences
Major Field of Study
First Term
Earned Degree
Did not Earn Degree
 
53 
 
 
Figure 68 
 
Figure 69 
With the exception of the Life_Partner 
sub-category (n = 7) where no students 
earned a degree, the J48 predictions 
were off by about 2%. 
 
Figure 70 
 
Figure 71 
For the students that were financially 
dependent upon their parents (≈69% of 
the cohort), J48 came within 0.6% of the 
actual outcome.  The Earned Degree 
predictions for the remaining 
subcategories, each with fewer instances,  
were lower than the actual outcomes by 
approximately 4%. 
 
14.29%
22.73%
31.79%
85.71%
77.27%
68.21%
0%
20%
40%
60%
80%
100%
Marital Status
Predicted - Yes
Predicted - No
20.45%
33.77%
100%
79.55%
66.23%
0%
20%
40%
60%
80%
100%
Marital Status
Earned Degree
Did not Earn 
Degree
35.33%
8.79%
29.05%
64.67%
91.21%
70.95%
0%
20%
40%
60%
80%
100%
Financially Dependent Upon Parents
Predicted - Yes
Predicted - No
35.98%
12.64%
33.70%
64.02%
87.36%
66.30%
0%
20%
40%
60%
80%
100%
Financially Dependent Upon Parents
Earned Degree
Did not Earn 
Degree
 
54 
 
 
Figure 72 
 
Figure 73 
J48’s predictions were between 1-2% 
different than the actual outcomes with 
the exceptions of $10,001-$12,000 and 
$16,001-$18,000, which were 5% and 
10% different respectively. 
Note:  505, or 25% of the cohort either did not 
complete or did not have a valid FAFSA 
required for determination of most financial 
awards. 
 
Figure 74 
 
Figure 75 
9-Month estimated expected family 
contribution predictions were fairly 
accurate for those students expected to 
pay $3,001 or more for their education 
than for those expected to pay less.   
 
 
37.28%
25.53%
39.33%
22.50%
62.72%
74.47%
60.67%
77.50%
0%
20%
40%
60%
80%
100%
Cost of Attendance
(For those with valid FAFSA)
Predicted - Yes
Predicted - No
38.90%
26.76%
37.08%
32.50%
61.10%
73.24%
62.92%
67.50%
0%
20%
40%
60%
80%
100%
Cost of Attendance
(For those with valid FAFSA)
Earned Degree
Did not Earn 
Degree
21.84%
41.38%
78.16%
58.62%
0%
20%
40%
60%
80%
Less than 
$3,001
$3,001 or 
Greater
9-Month Estimated Expected Family 
Contribution
(For those with valid FAFSA)
Predicted - Yes
Predicted - No
25.56%
40.43%
74.44%
59.57%
0%
20%
40%
60%
80%
Less than 
$3,001
$3,001 or 
Greater
9-Month Estimated Expected Family 
Contribution
(For those with valid FAFSA)
Earned Degree
Did not Earn 
Degree
 
55 
 
 
Figure 76 
 
Figure 77 
For the most part, the predicted and 
actual outcome charts are very similar.  
The findings for this category were 
consistent with common understanding 
that those who are more affluent have a 
greater tendency to graduate within the 
normal expected amount of time. 
 
 
Figure 78 
 
Figure 79 
Consistent with sub-categories 
containing a large percentage of the 
cohort, the predictions for those students 
receiving any financial aid were closer to 
the mark than those receiving no 
financial aid. 
 
 
0%
20%
40%
60%
80%
100%
Need Level
(For those with valid FAFSA)
Predicted - Yes
Predicted - No
0%
20%
40%
60%
80%
100%
Need Level
(For those with valid FAFSA)
Earned Degree
Did not Earn 
Degree
31.63%
24.00%
68.37%
76.00%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Any Aid - Yes Any Aid - No
Received Any Financial Aid
Predicted - Yes
Predicted - No
33.58%
16.00%
66.42%
84.00%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Any Aid - Yes Any Aid - No
Received Any Financial Aid
Earned Degree
Did not Earn 
Degree
 
56 
 
 
Figure 80 
 
Figure 81 
The accuracy of the predictions ranged 
between 1% and 4.5% different than the 
actual outcomes. 
 
 
Figure 82 
 
Figure 83 
In all sub-categories a greater percentage 
of students earned degrees than was 
predicted with the exception of the one 
student that fell in the Over $2,500 
range. 
 
 
 
41.29%
33.99%
19.08%
58.71%
66.01%
80.92%
0%
20%
40%
60%
80%
100%
No Aid Between 
$1 and 
$999
$1,000 
and Over
Federal Financial Aid
(For those with valid FAFSA)
Predicted - Yes
Predicted - No
40.28%
38.42%
22.16%
59.72%
61.58%
77.84%
0%
20%
40%
60%
80%
100%
No Aid Between 
$1 and 
$999
$1,000 
and Over
Federal Financial Aid
(For those with valid FAFSA)
Earned Degree
Did not Earn 
Degree
39.21%
25.74%
20.86%
20.45%
60.79%
74.26%
79.14%
79.55%
100% 100%
0%
20%
40%
60%
80%
100%
State Aid Ranges
(For those with valid FAFSA)
Predicted - Yes
Predicted - No
39.10%
28.71%
22.09%
25.00%
14.29%
60.90%
71.29%
77.91%
75.00%
85.71%
100%
0%
20%
40%
60%
80%
100%
State Aid Ranges
(For those with valid FAFSA)
Earned Degree
Did not Earn 
Degree
 
57 
 
 
Figure 84 
 
Figure 85 
Slightly more students in both sub-
categories earned degrees than were 
predicted. 
 
 
Figure 86 
 
Figure 87 
Over 4.5% more students in the no 
institutional aid range earned degree 
than were predicted.  In comparison 
between 2.2 and 2.8% fewer students 
earned degrees in the remaining ranges. 
 
 
 
32.66%
34.21%
67.34%
65.79%
0%
20%
40%
60%
80%
No Work 
Study Aid
Awarded 
Work Study 
Aid
Federal Work Study Aid Recipients 
(For those with valid FAFSA)
Predicted - Yes
Predicted - No
33.74%
36.84%
66.26%
63.16%
0%
20%
40%
60%
80%
No Work 
Study Aid
Awarded Work 
Study Aid
Federal Work Study Aid Recipients 
(For those with valid FAFSA)
Earned Degree
Did not Earn 
Degree
18.01%
44.50%
62.64%
81.99%
55.50%
37.36%
0%
20%
40%
60%
80%
100%
Institutional Aid Ranges
Predicted - Yes
Predicted - No
22.62%
41.63%
60.44%
77.38%
58.37%
39.56%
0%
20%
40%
60%
80%
100%
Institutional Aid Ranges
Earned Degree
Did not Earn 
Degree
 
58 
 
 
Figure 88 
 
Figure 89 
Other third party aid prediction 
percentages differed by 3%, 2.3%, and 
1.44% respectively. 
 
 
Figure 90 
 
Figure 91 
Student loan ranges prediction 
percentages differed between 1.39% and 
5.88% with the exception of the $8,001 
to $8,500 group which was exactly 
precise. 
 
 
 
10.53%
30.27%
48.92%
89.47%
69.73%
51.08%
0%
20%
40%
60%
80%
100%
No Third 
Party Aid
Between 
$1 and 
$600
$601 or 
More
Other Third Party Aid
Predicted - Yes
Predicted - No
13.53%
32.57%
47.48%
86.47%
67.43%
52.52%
0%
20%
40%
60%
80%
100%
No Third 
Party Aid
Between 
$1 and 
$600
$601 or 
More
Other Third Party Aid
Earned Degree
Did not Earn 
Degree
36.34%
29.77%
15.35%
25.32%
29.41%
33.33%
63.66%
70.23%
84.65%
74.68%
70.59%
66.67%
0%
20%
40%
60%
80%
100%
Student Loan Ranges
Predicted - Yes
Predicted - No
37.73%
31.32%
19.69%
29.11%
23.53%
33.33%
62.27%
68.68%
80.31%
70.89%
76.47%
66.67%
0%
20%
40%
60%
80%
100%
Student Loan Ranges
Earned Degree
Did not Earn 
Degree
 
59 
 
 
Figure 92 
 
Figure 93 
J48 performed well in predicting the 
percentage of full-time students who 
would later earn their degree but showed 
signs of difficulty with the part-time 
students’ prediction.  Again it may be 
worth stating that 1,900 of 2,020 
students in the cohort attended full-time 
their first term.  
 
Figure 94 
 
Figure 95 
In regard to the first term attempted 
credit hours sub-category, with the 
exception of the group of students 
attempting earn no credit hours their first 
term, J48 did not perform as well as 
expected.  Once again this may be an 
affect of the missing transfer out 
information. 
 
33.37%
2.50%
66.63%
97.50%
0%
20%
40%
60%
80%
100%
Full-Time Part-Time
First Term Academic Load
Predicted - Yes
Predicted - No
34.74%
11.67%
65.26%
88.33%
0%
20%
40%
60%
80%
100%
Full-Time Part-Time
First Term Academic Load
Earned Degree
Did not Earn 
Degree
1.15%
36.33%
70.23%
100% 100%
98.85%
63.67%
29.77%
0%
20%
40%
60%
80%
100%
0 1-5 6-10 11-1516-19
First Term Attempted Credit Hours
Predicted - Yes
Predicted - No
2.38%
9.22%
39.40%
58.40%
100%
97.62%
90.78%
60.60%
41.60%
0%
20%
40%
60%
80%
100%
0 1-5 6-10 11-15 16-19
First Term Attempted Credit Hours
Earned Degree
Did not Earn 
Degree
 
60 
 
 
Figure 96 
 
Figure 97 
J48 performed well in predicting the 
percentage of students earning between 
12 and 16 credit hours that would later 
earn their degree but showed a problem 
with the 17.00+ credit hour group’s 
prediction.  This may well be another 
attribute affected by the missing transfer 
out information.    
 
 
Figure 98 
 
Figure 99 
J48’s predictions for students ending 
their first term with 13-24, 25-36, 49-60, 
and 61-76 quality points were 
significantly different than the actual 
values.  This attribute also falls on the 
list of casualties with regard to transfer 
out information.  
1.51%
48.50%
66.47%
100%
98.49%
51.50%
33.53%
0%
20%
40%
60%
80%
100%
No 
Hours
1.00 to 
11.00
12.00 to 
16.00
17.00+
Credit Hours Earned
Predicted - Yes
Predicted - No
10.21%
49.16%
56.89%
100%
89.79%
50.84%
43.11%
0%
20%
40%
60%
80%
100%
No Hours 1.00 to 
11.00
12.00 to 
16.00
17.00+
Credit Hours Earned
Earned Degree
Did not Earn 
Degree
0.99%
8.55%
55.20%
87.54%
84.51%
100% 100%
99.01%
91.45%
44.80%
12.46%
15.49%
0%
20%
40%
60%
80%
100%
First Term Total Quality Points Earned
Predicted - Yes
Predicted - No
2.56%
16.50%
27.72%
50.96%
67.29%
78.87%
100%
97.44%
83.50%
72.28%
49.04%
32.71%
21.13%
0%
20%
40%
60%
80%
100%
First Term Total Quality Points Earned
Earned Degree
Did not Earn 
Degree
 
61 
 
 
Figure 100 
 
Figure 101 
Once more the students intuitively 
expected to earn degrees within the six-
year time period in actuality graduated in 
lesser percentages than predicted.  
Furthermore those students with lower 
first term GPAs graduated at a 
significantly higher rate.  This latter 
issue is an area worthy of follow-up 
investigation. 
 
Figure 102 
 
Figure 103 
J48’s prediction for those students with 
no remediation and for those with 
remediation was 2% and 5% different 
than the actual values respectively.  The 
counter-intuitive result in regard to those 
engaging in remediation is also an item 
that should be further explored.  Perhaps 
the introduction of an additional dataset 
would help increase the accuracy. 
0.73%
7.22%
34.19%
43.53%
66.49%
71.92%
78.85%
100%
99.27%
92.78%
65.81%
56.47%
33.51%
28.08%
21.15%
0%
20%
40%
60%
80%
100%
First Term GPA
Predicted - Yes
Predicted - No
13.55%
22.68%
39.03%
40.95%
57.84%
60.27%
70.48%
100%
86.45%
77.32%
60.97%
59.05%
42.16%
39.73%
29.52%
0%
20%
40%
60%
80%
100%
First Term GPA
Earned Degree
Did not Earn 
Degree
42.10%
22.96%
57.90%
77.04%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
No 
Remediation
Remediation
Any Remediation
Predicted - Yes
Predicted - No
40.11%
27.89%
59.89%
72.11%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
No 
Remediation
Remediation
Any Remediation
Earned Degree
Did not Earn 
Degree
 
62 
 
 
Figure 104 
 
Figure 105 
J48 did well in predicting the outcomes 
of those students engaging in no 
remedial English and those failing 
remedial English.  However it did not do 
well predicting the outcomes of those 
passing remedial English.  
 
Figure 106 
 
Figure 107 
Similarly J48’s performed well in 
predicting the percent of the cohort that 
would earn a degree for those students 
with no remedial mathematics and for 
those failing remedial mathematics.  
However, once again its prediction for 
those passing the remedial coursework 
was off by 5%.     
36.86%
0.97%
25.86%
63.14%
99.03%
74.14%
0%
20%
40%
60%
80%
100%
No 
Remedial 
English
Failed 
Remedial 
English
Passed 
Remedial 
English
Remedial English
Predicted - Yes
Predicted - No
35.69%
34.11%
64.31%
100%
65.89%
0%
20%
40%
60%
80%
100%
No 
Remedial 
English
Failed 
Remedial 
English
Passed 
Remedial 
English
Remedial English
Earned Degree
Did not Earn 
Degree
37.69%
0.46%
29.57%
62.31%
99.54%
70.43%
0%
20%
40%
60%
80%
100%
No 
Remedial 
Math
Failed 
Remedial 
Math
Passed 
Remedial 
Math
Remedial Mathematics
Predicted - Yes
Predicted - No
37.77%
2.29%
35.59%
62.23%
97.71%
64.41%
0%
20%
40%
60%
80%
100%
No 
Remedial 
Math
Failed 
Remedial 
Math
Passed 
Remedial 
Math
Remedial Mathematics
Earned Degree
Did not Earn 
Degree
 
63 
 
 
Figure 108 
 
Figure 109 
As was the case with the other remedial 
areas, J48 performed well on its 
predictions for those students not 
engaging in Reading & Study Skills 
coursework and those failing the 
coursework.  Yet its prediction for those 
passing the coursework was significantly 
different.  In this case J48 was off by 
over 8 percentage points.  
 
Figure 110 
 
Figure 111 
The predictions for those students never 
visiting the Center for Student Progress 
during their first term were very close to 
the actual percentages yet the predicted 
percentages were off by 6 percentage 
points for those that did visit the Center. 
 
 
34.70%
22.57%
65.30%
100%
77.43%
0%
20%
40%
60%
80%
100%
No 
Reading 
& Study 
Skills
Failed 
Reading 
& Study 
Skills
Passed 
Reading 
& Study 
Skills
Reading & Study Skills
Predicted - Yes
Predicted - No
35.08%
30.71%
64.92%
100%
69.29%
0%
20%
40%
60%
80%
100%
No 
Reading 
& Study 
Skills
Failed 
Reading 
& Study 
Skills
Passed 
Reading 
& Study 
Skills
Reading & Study Skills
Earned Degree
Did not Earn 
Degree
52.92%
27.69%
47.08%
72.31%
0%
20%
40%
60%
80%
100%
Visted the 
Center for 
Student 
Progress
Never Visited 
the Center for 
Student 
Progress
Center for Student Progress
Predicted - Yes
Predicted - No
56.82%
29.15%
43.18%
70.85%
0%
20%
40%
60%
80%
100%
Visted the 
Center for 
Student 
Progress
Never Visited 
the Center for 
Student 
Progress
Center for Student Progress
Earned Degree
Did not Earn 
Degree
 
64 
 
 
Figure 112 
 
Figure 113 
J48 did a stellar job predicting the 
percentage of students not enrolling the 
immediately following spring term that 
would go on to earn a degree and was 
off by a little more than 2% for your 
enrolling that spring term.  
 
Figure 114 
 
Figure 115 
Likewise, J48 performed well in 
predicting the percentages of students 
earning a degree with six years for both 
those students consecutively enrolled 
fall, spring and the following fall terms 
and for those not consecutively enrolled. 
36.79%
1.97%
63.21%
98.03%
0%
20%
40%
60%
80%
100%
Spring - Yes Spring - No
Returned Spring Term
Predicted - Yes
Predicted - No
39.01%
1.64%
60.99%
98.36%
0%
20%
40%
60%
80%
100%
Spring - Yes Spring - No
Returned Spring Term
Earned Degree
Did not Earn 
Degree
45.92%
1.97%
54.08%
98.03%
0%
20%
40%
60%
80%
100%
Spring & Fall - Yes Spring & Fall - No
Returned Spring and Next Fall
Predicted - Yes
Predicted - No
48.12%
3.03%
51.88%
96.97%
0%
20%
40%
60%
80%
100%
Spring & Fall - Yes Spring & Fall - No
Returned Spring and Next Fall
Earned Degree
Did not Earn 
Degree
 
65 
 
 
Figure 116
 
Figure 117 
J48 also did well predicting the 
percentage of students enrolling and not 
enrolling the fall term for the second 
year who would go on to earn a degrees 
– with predictions within a little more 
than 2% of the actual amounts.  
 
The table of statistics detailing the cohort distributions among the various data elements 
including the predicted and actual values can be found in Table 1. 
 
 
 
 
 
 
45.28%
1.27%
54.72%
98.73%
0%
20%
40%
60%
80%
100%
Next Fall -
Yes
Next Fall -
No
Returned the Next Fall
Predicted - Yes
Predicted - No
47.23%
2.85%
52.77%
97.15%
0%
20%
40%
60%
80%
100%
Next Fall -
Yes
Next Fall - No
Returned the Next Fall
Earned Degree
Did not Earn 
Degree
 
66 
 
4.  Discussion 
 With few exceptions, the J48 predictions were very close to the actual outcomes 
experienced by the students in the cohort.  Noticeable differences in the high school 
grade point average, first term grade point average, first term quality points earned, and 
first term credit hours may be explained by missing data – in particular data indicating 
whether or not students transferred out to another institution.  Introduction of this missing 
data may support the statements of Bowen, Chingos & McPherson (2009) with regard to 
the predictive strength of high school grade point average of student degree attainment.  
Additionally the 86.29% accuracy rate provides a strong support for future utilization of 
data mining on student data for success prediction.  In general the ease of use of the data 
mining software combined with the high rate of accuracy, make this method of prediction 
highly desirable.  By allowing the software to perform the difficult computations, the 
researcher was able to focus on those elements of the process most familiar - selecting 
appropriate student data attributes and preparing the dataset for processing.   
5.  Conclusions 
          Data mining software provides a relatively easy way to quickly identify previously 
unknown relationships among the attributes within a student cohort dataset.  These 
relationships may provide policy analysts with the necessary information for supporting 
operational changes in order to enhance a higher education institution’s graduation rate.  
Thusly by increasing the number of college credentialed citizens within the state, a higher 
education institution will provide a desirable educated workforce to entice new business 
and industry to the region.  As the economy has reached a significant low point and 
 
67 
 
unemployment rates continue to climb though at a decreasing rate, the use of such 
predictions can have a dramatic affect on the institution’s ability to provide outstanding 
service to the state as well as the students who enter its domain.  Further the resulting 
changes may reduce the time to degree and subsequently the cost of higher education to 
the student while increasing the institution’s subsidy allocation from the state. 
5.1  Recommendations 
          Follow-up work is indicated for this study.  The decision tree model produced with 
this dataset should be applied to future datasets to gauge and/or increase its accuracy and 
in all likelihood refine the decision tree model itself for subsequent use.  Additionally, 
further analysis of the cohort dataset including the J48 predictions is necessary for 
developing profiles of each student category for communicating to high school guidance 
counselors, for effective institutional recruiting efforts, for academic advising and for 
identifying appropriate intervention.  Moreover the introduction of transfer out data and 
possibly the expansion of the dataset to include teaching faculty attributes should be 
strongly considered in order to provide a stronger predictive result from the algorithm.  
Finally investigation of the counter-intuitive results with regard to remedial coursework 
should be explored.  
 
 
 
 
 
 
68 
 
REFERENCES 
Bailey, B.  (2006). Let the data talk:  developing models to explain IPEDS graduation  
rates.    New Directions for Institutional Research n 131 p 101-115 fall 2006.   
Bowen, W., Chingos, M., & McPherson, M.  (2009, October 16).  Crossing the finish line 
completing at America’s public universities.  Presentation to the Ohio Association 
for Institutional Research and Planning fall conference 2009.  Columbus, Ohio:  
The Ohio State University. 
Center for Student Progress.  Mission statement.  Retrieved October 17, 2009, from 
http://www.ysu.edu/csp/historymission.shtml. 
Executive Office of the President, Council of Economic Advisors. (2009, July).  
Preparing the workers of today for the jobs of tomorrow.  Retrieved September  
20, 2009, from http://www.whitehouse.gov/administration/eop/cea/Jobs-of-the-
Future/. 
Herzog, S. (2006, Fall).  Estimating student retention and degree-completion time:  
decision trees and neural networks vis-à-vis regression.  New Directions for 
Institutional Research, no 131. 
Jacobs, L. & Hyman, J. (2009, July 1).  7 reasons why college is so expensive.  U.S.  
 News & World Report, Professors Guide.  Retrieved July 9, 2009, from 
http://www.usnews.com/blogs/professors-guide.                        
Jaschik, S. (2008, October 9).  Falling behind.  Inside Higher Ed.  Retrieved October 11, 
2008, from http://insidehighered.com/news/2008/10/09/minority. 
Long, W., Griffith, J., Selker, H., & D’Agostino, R.  (1993).  A comparison of logistic 
regression to decision-tree induction in a medical domain” from Computers in  
 
69 
 
Biomedical Research, 26: 74-97, 1993. 
 
Moltz, D. (2009, April 30).  Adopting performance-based funding.  Inside Higher 
Education.  Retrieved May 1, 2009, from 
http://www.insidehighered.com/layout/set/print/news/2009/04/30/ohio 5/1/2009. 
Ohio Board of Regents. (2009, November).  Higher Education Information System,  
Mission and Purpose.  Retrieved November 15, 2009, from  
http://regents.ohio.gov/hei/.  
Perry, N. (2008, October 6, 2008).  With no way out of trouble, more students likely to 
default.  The Seattle Times.  Retrieved October 11, 2008, from 
http://seattletimes.nwsource.com/html/localnews/2008231488_loandaytwo06m.ht
ml. 
Roe, B., Yang, H., Zhu, J., Liu, Y., Stancu, I., & McGregor, G.  (2005).  Boosted 
decision trees as an alternative to artificial neural networks for particle 
identification.  Nuclear Instruments and Methods in Physics Research A 543 
(2005) 577-584. 
Swail, W. (2008, August 22).  The bell curve under a different cover.  The Educational  
Policy Institute’s – Week in Review.  Retrieved August 30, 2009, from 
http://www.educationalpolicy.org/pub/wir/080822.html. 
Tan, P., Steinbach, M., & Kumar, V. (2006).  Introduction to Data Mining.  Pearson  
 Education, Inc.  Upper Saddle River, New Jersey.   
The University of Waikato.  (2009).  Weka machine learning project.  Retrieved October  
17, 2009. from http://www.cs.waikato.ac.nz/~ml/index.html. 
 
70 
 
Witten, I. & Frank, E.  Data mining practical machine learning tools and techniques (2
nd 
ed.).  Elsevier, Inc.  Maryland Heights, Missouri. 
Youngstown State University.  Undergraduate Student Bulletin.  Courses, Reading &  
Study Skills.  Retrieved October 19, 2009, from 
http://www.ysu.edu/catalog/files/Courses%20251-401.pdf , p.384. 
 
 
Appendix A 
=== Run information === 
 
Scheme:       weka.classifiers.trees.J48 -C 0.25 -M 2 
Relation:     2001 discret. no-yes no  pids Pick ME csv-weka.filters.unsupervised.attribute.Remove-R37-
40 
Instances:    2020 
Attributes:   36 
              PID 
              SP_Ret 
              AU_Next_Ret 
              HS_GPA_Range 
              Gender 
              Age_Range 
              Ethnicity 
              Academic_Intent 
              State_Resident 
              Cum_GPA_CrHr_Ranges 
              Cum_QPts_Ranges 
              Cum_GPA_Ranges 
              Cum_Credit_Hour_Range 
              Major 
              Commuter 
              Fed_Aid_Excl_Loans_Range 
              State_Aid_Range 
              Work_Study_Range 
              Student_Loan_Range 
              Institutional_Aid_Range 
              Other_3rd_Party_Aid_Range 
              Dependency 
              Cost_Of_Attendance_Range 
              9_Month_Expected_Family_Contribution 
              Need_Level_Range 
              Student_Marital_Status 
              Load 
              Comp_ACT_Ranges 
              HS_CEEB_Code 
              Any_AP 
              Associate_Ever 
              Bachelor_Degree 
              #_of_CSP_Visits 
              Remedial_English 
              Remedial_Math 
              R&SK 
Test mode:    evaluate on training data 
 
=== Classifier model (full training set) === 
 
J48 pruned tree 
------------------ 
 
AU_Next_Ret = NO 
|   #_of_CSP_Visits = None: No (618.0/10.0) 
|   #_of_CSP_Visits = 1. 1-5 
|   |   Fed_Aid_Excl_Loans_Range = 9. 1750-1999: No (2.0) 
|   |   Fed_Aid_Excl_Loans_Range = 1. No_Aid: Yes (7.0) 
|   |   Fed_Aid_Excl_Loans_Range = 5. 750-999: Yes (1.0) 
|   |   Fed_Aid_Excl_Loans_Range = 7. 1250-1499: Yes (0.0) 
|   |   Fed_Aid_Excl_Loans_Range = 3. 200-499: Yes (0.0) 
|   |   Fed_Aid_Excl_Loans_Range = 4. 500-749: No (1.0) 
|   |   Fed_Aid_Excl_Loans_Range = 11. 2250-2499: No (1.0) 
|   |   Fed_Aid_Excl_Loans_Range = 6. 1000-1249: Yes (0.0) 
|   |   Fed_Aid_Excl_Loans_Range = 8. 1500-1749: Yes (0.0) 
|   |   Fed_Aid_Excl_Loans_Range = 2. 1-199: Yes (0.0) 
|   |   Fed_Aid_Excl_Loans_Range = 10. 2000-2249: Yes (0.0) 
|   |   Fed_Aid_Excl_Loans_Range = 12. 2500-2749: Yes (0.0) 
|   #_of_CSP_Visits = 5. 21+: No (1.0) 
|   #_of_CSP_Visits = 2. 6-10: No (0.0) 
|   #_of_CSP_Visits = 3. 11-15: No (0.0) 
|   #_of_CSP_Visits = 4. 16-20: No (0.0) 
AU_Next_Ret = YES 
|   Cum_QPts_Ranges = 2. 1-12: No (72.0/5.0) 
|   Cum_QPts_Ranges = 3. 13-24: No (194.0/43.0) 
|   Cum_QPts_Ranges = 5. 37-48 
|   |   Age_Range = 35-39: No (2.0) 
|   |   Age_Range = 30-34: Yes (4.0/1.0) 
|   |   Age_Range = 40-49: No (2.0) 
|   |   Age_Range = 25-29: No (4.0/1.0) 
|   |   Age_Range = 18-19 
|   |   |   Cum_Credit_Hour_Range = 7.00-11.00: No (4.0/1.0) 
|   |   |   Cum_Credit_Hour_Range = 1.00-6.00: Yes (0.0) 
|   |   |   Cum_Credit_Hour_Range = 12.00-16.00 
|   |   |   |   #_of_CSP_Visits = None 
|   |   |   |   |   Major = Health Professions and Clinical Services 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 2. 1-100: Yes (6.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 1. No_3rd_Party_Aid: No (3.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 7. 1501-2500: Yes (1.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 4. 501-600 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 9. 1750-1999: Yes (2.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 1. No_Aid: No (2.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 5. 750-999: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 7. 1250-1499: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 3. 200-499: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 4. 500-749: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 11. 2250-2499: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 6. 1000-1249: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 8. 1500-1749: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 2. 1-199: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 10. 2000-2249: No (0.0) 
|   |   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 12. 2500-2749: No (0.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 8. 2501-5000: Yes (0.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 6. 1001-1500: Yes (0.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 5. 601-1000: Yes (0.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 3. 101-500 
|   |   |   |   |   |   |   HS_CEEB_Code <= 361870: Yes (2.0) 
|   |   |   |   |   |   |   HS_CEEB_Code > 361870: No (2.0) 
|   |   |   |   |   |   Other_3rd_Party_Aid_Range = 9. 5001-8000: Yes (0.0) 
|   |   |   |   |   Major = Business Management and Marketing: Yes (38.0/9.0) 
|   |   |   |   |   Major = Public Administration and Social Service: Yes (3.0) 
|   |   |   |   |   Major = Computer and Information Sciences 
|   |   |   |   |   |   Ethnicity = Black: No (0.0) 
|   |   |   |   |   |   Ethnicity = White: No (4.0/1.0) 
|   |   |   |   |   |   Ethnicity = Unspecified_Race: Yes (2.0) 
|   |   |   |   |   |   Ethnicity = Hispanic: No (0.0) 
|   |   |   |   |   |   Ethnicity = International: No (0.0) 
|   |   |   |   |   |   Ethnicity = Asian: No (0.0) 
|   |   |   |   |   |   Ethnicity = American_Indian: No (0.0) 
|   |   |   |   |   Major = Social Sciences: Yes (4.0) 
|   |   |   |   |   Major = Education 
|   |   |   |   |   |   Remedial_Math = Failed: No (1.0) 
|   |   |   |   |   |   Remedial_Math = Did_not_take: Yes (28.0/6.0) 
|   |   |   |   |   |   Remedial_Math = Passed 
|   |   |   |   |   |   |   Institutional_Aid_Range = 1. No_Institutional_Aid: No (9.0/1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 2. 1-500 
|   |   |   |   |   |   |   |   HS_CEEB_Code <= 365507: Yes (3.0) 
|   |   |   |   |   |   |   |   HS_CEEB_Code > 365507: No (2.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 6. 2001-3000: No (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 7. 3001-4000: No (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 3. 501-1000: Yes (1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 4. 1001-1500: No (1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 9. 5001-6000: No (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 5. 1501-2000: No (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range =  10. 6001-7000: No (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range =  11. 7001-8000: No (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 8. 4001-5000: No (0.0) 
|   |   |   |   |   Major = Biological and Biomedical Sciences 
|   |   |   |   |   |   Academic_Intent = Obtain_Bachelors_Degree: Yes (2.0) 
|   |   |   |   |   |   Academic_Intent = Obtain_Associate_Degree_for_Job_Market: Yes (0.0) 
|   |   |   |   |   |   Academic_Intent = Personal_Interest: No (1.0) 
|   |   |   |   |   |   Academic_Intent = Unknown 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 2. 1-100: Yes (2.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 1. No_3rd_Party_Aid: No (0.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 7. 1501-2500: No (0.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 4. 501-600: No (2.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 8. 2501-5000: No (0.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 6. 1001-1500: No (0.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 5. 601-1000: No (0.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 3. 101-500: No (0.0) 
|   |   |   |   |   |   |   Other_3rd_Party_Aid_Range = 9. 5001-8000: No (0.0) 
|   |   |   |   |   |   Academic_Intent = Selected_Courses_Train_New_Career: Yes (0.0) 
|   |   |   |   |   |   Academic_Intent = Selected_Courses_Upgrade_Skills: Yes (0.0) 
|   |   |   |   |   |   Academic_Intent = Transfer_Before_Degree: Yes (0.0) 
|   |   |   |   |   |   Academic_Intent = Obtain_Associate_Degree_for_Transfer: Yes (0.0) 
|   |   |   |   |   |   Academic_Intent = Obtain_Undergraduate_Certificate: Yes (0.0) 
|   |   |   |   |   Major = Engineering 
|   |   |   |   |   |   Ethnicity = Black: Yes (0.0) 
|   |   |   |   |   |   Ethnicity = White: Yes (16.0/2.0) 
|   |   |   |   |   |   Ethnicity = Unspecified_Race: No (3.0/1.0) 
|   |   |   |   |   |   Ethnicity = Hispanic: Yes (0.0) 
|   |   |   |   |   |   Ethnicity = International: Yes (0.0) 
|   |   |   |   |   |   Ethnicity = Asian: Yes (0.0) 
|   |   |   |   |   |   Ethnicity = American_Indian: No (1.0) 
|   |   |   |   |   Major = Liberal Arts and General Studies 
|   |   |   |   |   |   Ethnicity = Black: No (0.0) 
|   |   |   |   |   |   Ethnicity = White 
|   |   |   |   |   |   |   HS_CEEB_Code <= 363487: No (6.0) 
|   |   |   |   |   |   |   HS_CEEB_Code > 363487 
|   |   |   |   |   |   |   |   State_Resident = Yes: Yes (22.0/6.0) 
|   |   |   |   |   |   |   |   State_Resident = No: No (2.0) 
|   |   |   |   |   |   Ethnicity = Unspecified_Race: No (3.0) 
|   |   |   |   |   |   Ethnicity = Hispanic: No (0.0) 
|   |   |   |   |   |   Ethnicity = International: No (0.0) 
|   |   |   |   |   |   Ethnicity = Asian: No (0.0) 
|   |   |   |   |   |   Ethnicity = American_Indian: No (0.0) 
|   |   |   |   |   Major = Security and Protective Services 
|   |   |   |   |   |   Gender = Female: Yes (4.0) 
|   |   |   |   |   |   Gender = Male: No (4.0/1.0) 
|   |   |   |   |   Major = Natural Resources and Conservation: Yes (0.0) 
|   |   |   |   |   Major = English Language and Literature: No (6.0/1.0) 
|   |   |   |   |   Major = Area, Ethnic, Cultural, Gender Studies: Yes (0.0) 
|   |   |   |   |   Major = Visual and Performing Arts 
|   |   |   |   |   |   HS_CEEB_Code <= 365310: No (11.0) 
|   |   |   |   |   |   HS_CEEB_Code > 365310: Yes (11.0/4.0) 
|   |   |   |   |   Major = Legal Professions and Studies 
|   |   |   |   |   |   PID <= 250530: No (2.0) 
|   |   |   |   |   |   PID > 250530: Yes (3.0) 
|   |   |   |   |   Major = Foreign Languages and Literature: Yes (0.0) 
|   |   |   |   |   Major = Engineering Technology: No (1.0) 
|   |   |   |   |   Major = Psychology 
|   |   |   |   |   |   Gender = Female: No (9.0/2.0) 
|   |   |   |   |   |   Gender = Male: Yes (3.0/1.0) 
|   |   |   |   |   Major = Physical Sciences: Yes (7.0/1.0) 
|   |   |   |   |   Major = Mathematics and Statistics: No (1.0) 
|   |   |   |   |   Major = Communication and Journalism: No (3.0/1.0) 
|   |   |   |   |   Major = Precision Production: Yes (1.0) 
|   |   |   |   |   Major = Leisure and Fitness Studies 
|   |   |   |   |   |   Remedial_Math = Failed: No (0.0) 
|   |   |   |   |   |   Remedial_Math = Did_not_take 
|   |   |   |   |   |   |   Remedial_English = Passed: No (2.0) 
|   |   |   |   |   |   |   Remedial_English = Did_not_take: Yes (7.0) 
|   |   |   |   |   |   |   Remedial_English = Failed: Yes (0.0) 
|   |   |   |   |   |   Remedial_Math = Passed: No (5.0) 
|   |   |   |   |   Major = Family and Consumer Sciences: Yes (3.0/1.0) 
|   |   |   |   |   Major = Philosophy and Religious Studies: Yes (0.0) 
|   |   |   |   #_of_CSP_Visits = 1. 1-5 
|   |   |   |   |   State_Aid_Range = 3. 501-1000: Yes (5.0/2.0) 
|   |   |   |   |   State_Aid_Range = 2. 1-500: Yes (11.0/1.0) 
|   |   |   |   |   State_Aid_Range = 4. 1001-1500: No (4.0/1.0) 
|   |   |   |   |   State_Aid_Range = 1. No_State_Aid 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 9. 1750-1999: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 1. No_Aid 
|   |   |   |   |   |   |   Institutional_Aid_Range = 1. No_Institutional_Aid 
|   |   |   |   |   |   |   |   Dependency = I: Yes (0.0) 
|   |   |   |   |   |   |   |   Dependency = D: Yes (16.0/3.0) 
|   |   |   |   |   |   |   |   Dependency =  : No (11.0/3.0) 
|   |   |   |   |   |   |   |   Dependency = X: Yes (0.0) 
|   |   |   |   |   |   |   |   Dependency = Y: Yes (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 2. 1-500: Yes (17.0/4.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 6. 2001-3000: No (1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 7. 3001-4000: No (1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 3. 501-1000 
|   |   |   |   |   |   |   |   PID <= 251467: Yes (2.0) 
|   |   |   |   |   |   |   |   PID > 251467: No (2.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 4. 1001-1500: No (2.0/1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 9. 5001-6000: Yes (1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 5. 1501-2000: No (1.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range =  10. 6001-7000: Yes (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range =  11. 7001-8000: Yes (0.0) 
|   |   |   |   |   |   |   Institutional_Aid_Range = 8. 4001-5000: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 5. 750-999: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 7. 1250-1499: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 3. 200-499: No (2.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 4. 500-749: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 11. 2250-2499: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 6. 1000-1249: Yes (1.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 8. 1500-1749: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 2. 1-199: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 10. 2000-2249: Yes (0.0) 
|   |   |   |   |   |   Fed_Aid_Excl_Loans_Range = 12. 2500-2749: Yes (0.0) 
|   |   |   |   |   State_Aid_Range = 6. 2001-2500: Yes (0.0) 
|   |   |   |   |   State_Aid_Range = 7. Over 2500: Yes (0.0) 
|   |   |   |   #_of_CSP_Visits = 5. 21+: Yes (0.0) 
|   |   |   |   #_of_CSP_Visits = 2. 6-10: Yes (9.0) 
|   |   |   |   #_of_CSP_Visits = 3. 11-15: No (4.0/2.0) 
|   |   |   |   #_of_CSP_Visits = 4. 16-20: Yes (2.0) 
|   |   |   Cum_Credit_Hour_Range = 0.0: Yes (0.0) 
|   |   |   Cum_Credit_Hour_Range = 22.00-26.00: Yes (3.0/1.0) 
|   |   |   Cum_Credit_Hour_Range = 17.00-21.00 
|   |   |   |   Academic_Intent = Obtain_Bachelors_Degree 
|   |   |   |   |   PID <= 251102: Yes (11.0/2.0) 
|   |   |   |   |   PID > 251102: No (4.0) 
|   |   |   |   Academic_Intent = Obtain_Associate_Degree_for_Job_Market: No (0.0) 
|   |   |   |   Academic_Intent = Personal_Interest: No (1.0) 
|   |   |   |   Academic_Intent = Unknown: No (9.0/1.0) 
|   |   |   |   Academic_Intent = Selected_Courses_Train_New_Career: No (0.0) 
|   |   |   |   Academic_Intent = Selected_Courses_Upgrade_Skills: No (0.0) 
|   |   |   |   Academic_Intent = Transfer_Before_Degree: No (1.0) 
|   |   |   |   Academic_Intent = Obtain_Associate_Degree_for_Transfer: No (0.0) 
|   |   |   |   Academic_Intent = Obtain_Undergraduate_Certificate: No (0.0) 
|   |   |   Cum_Credit_Hour_Range = 27.00-31.00: Yes (0.0) 
|   |   |   Cum_Credit_Hour_Range = 32.00-over: No (3.0/1.0) 
|   |   Age_Range = 22-24 
|   |   |   Comp_ACT_Ranges = 2. 6-11: No (0.0) 
|   |   |   Comp_ACT_Ranges = No ACT: No (8.0/2.0) 
|   |   |   Comp_ACT_Ranges = 4. 18-23: No (0.0) 
|   |   |   Comp_ACT_Ranges = 5. 24-29: Yes (2.0) 
|   |   |   Comp_ACT_Ranges = 6. 30-36: No (0.0) 
|   |   |   Comp_ACT_Ranges = 3. 12-17: No (0.0) 
|   |   Age_Range = 50-64: Yes (0.0) 
|   |   Age_Range = 20-21 
|   |   |   9_Month_Expected_Family_Contribution = 1. No_Family_Contribution: No (2.0) 
|   |   |   9_Month_Expected_Family_Contribution = 5. 5001-7000: Yes (0.0) 
|   |   |   9_Month_Expected_Family_Contribution = 9. 13001-20000: Yes (0.0) 
|   |   |   9_Month_Expected_Family_Contribution = 2. 1-1000: Yes (0.0) 
|   |   |   9_Month_Expected_Family_Contribution = 6. 7001-9000: Yes (0.0) 
|   |   |   9_Month_Expected_Family_Contribution = 4. 3001-5000: Yes (2.0) 
|   |   |   9_Month_Expected_Family_Contribution = 10. 20001-40000: Yes (0.0) 
|   |   |   9_Month_Expected_Family_Contribution = 3. 1001-3000: Yes (1.0) 
|   |   |   9_Month_Expected_Family_Contribution = 7. 9001-11000: Yes (0.0) 
|   |   |   9_Month_Expected_Family_Contribution = 8. 11001-13000: Yes (0.0) 
|   |   |   9_Month_Expected_Family_Contribution = 11. Over 40001: Yes (0.0) 
|   |   Age_Range = Under_18: Yes (0.0) 
|   Cum_QPts_Ranges = 4. 25-36 
|   |   #_of_CSP_Visits = None: No (238.0/68.0) 
|   |   #_of_CSP_Visits = 1. 1-5 
|   |   |   Ethnicity = Black: No (2.0) 
|   |   |   Ethnicity = White 
|   |   |   |   Remedial_English = Passed 
|   |   |   |   |   Cum_GPA_CrHr_Ranges = 3. 6-10: No (7.0/1.0) 
|   |   |   |   |   Cum_GPA_CrHr_Ranges = 4. 11-15 
|   |   |   |   |   |   Commuter = Yes: No (14.0/6.0) 
|   |   |   |   |   |   Commuter = No: Yes (3.0) 
|   |   |   |   |   Cum_GPA_CrHr_Ranges = 2. 1-5: No (0.0) 
|   |   |   |   |   Cum_GPA_CrHr_Ranges = 1. 0: No (0.0) 
|   |   |   |   |   Cum_GPA_CrHr_Ranges = 5. 16-19: Yes (1.0) 
|   |   |   |   Remedial_English = Did_not_take: Yes (16.0/2.0) 
|   |   |   |   Remedial_English = Failed: Yes (0.0) 
|   |   |   Ethnicity = Unspecified_Race: No (3.0/1.0) 
|   |   |   Ethnicity = Hispanic: No (2.0) 
|   |   |   Ethnicity = International: Yes (0.0) 
|   |   |   Ethnicity = Asian: No (1.0) 
|   |   |   Ethnicity = American_Indian: Yes (0.0) 
|   |   #_of_CSP_Visits = 5. 21+: No (1.0) 
|   |   #_of_CSP_Visits = 2. 6-10 
|   |   |   R&SK = Passed: Yes (3.0) 
|   |   |   R&SK = Did_not_take: No (4.0) 
|   |   |   R&SK = Failed: No (0.0) 
|   |   #_of_CSP_Visits = 3. 11-15 
|   |   |   State_Aid_Range = 3. 501-1000: Yes (0.0) 
|   |   |   State_Aid_Range = 2. 1-500: Yes (0.0) 
|   |   |   State_Aid_Range = 4. 1001-1500: No (2.0) 
|   |   |   State_Aid_Range = 1. No_State_Aid: Yes (7.0/1.0) 
|   |   |   State_Aid_Range = 6. 2001-2500: Yes (0.0) 
|   |   |   State_Aid_Range = 7. Over 2500: Yes (0.0) 
|   |   #_of_CSP_Visits = 4. 16-20: Yes (1.0) 
|   Cum_QPts_Ranges = 1. 0: No (56.0) 
|   Cum_QPts_Ranges = 6. 49-60 
|   |   Associate_Ever = No: Yes (280.0/69.0) 
|   |   Associate_Ever = Yes: No (8.0/2.0) 
|   Cum_QPts_Ranges = 7. 61-76 
|   |   Associate_Ever = No: Yes (60.0/4.0) 
|   |   Associate_Ever = Yes: No (2.0) 
 
Number of Leaves  :  227 
 
Size of the tree :  272 
 
 
Time taken to build model: 0.05 seconds 
 
 
=== Evaluation on training set === 
=== Summary === 
 
Correctly Classified Instances        1743               86.2871 % 
Incorrectly Classified Instances       277               13.7129 % 
Kappa statistic                          0.6873 
Mean absolute error                      0.2074 
Root mean squared error                  0.3221 
Relative absolute error                 46.6442 % 
Root relative squared error             68.3008 % 
Total Number of Instances             2020      
 
=== Detailed Accuracy By Class === 
 
               TP Rate   FP Rate   Precision   Recall  F-Measure   ROC Area  Class 
                 0.911     0.233      0.886     0.911     0.898      0.918    No 
                 0.767     0.089      0.812     0.767     0.789      0.918    Yes 
Weighted Avg.    0.863     0.185      0.862     0.863     0.862      0.918 
 
=== Confusion Matrix === 
 
    a    b   <-- classified as 
 1226  120 |    a = No 
  157  517 |    b = Yes 
 
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
Gender
Female 395 58.61% 674 50.07% 1,069 310 59.96% 72 60.00% 382 35.73% 85 54.14% 602 49.10% 687 64.27%
Male 279 41.39% 672 49.93% 951 207 40.04% 48 40.00% 255 26.81% 72 45.86% 624 50.90% 696 73.19%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Age Ranges
Under_18 1 0.15% 1 0.07% 2 0.00% 0.00% 0.00% 1 0.64% 1 0.08% 2 100.00%
18-19 642 95.25% 1,093 81.20% 1,735 500 96.71% 111 92.50% 611 35.22% 142 90.45% 982 80.10% 1,124 64.78%
20-21 9 1.34% 85 6.32% 94 6 1.16% 2 1.67% 8 8.51% 3 1.91% 83 6.77% 86 91.49%
22-24 8 1.19% 70 5.20% 78 3 0.58% 2 1.67% 5 6.41% 5 3.18% 68 5.55% 73 93.59%
25-29 7 1.04% 38 2.82% 45 3 0.58% 1 0.83% 4 8.89% 4 2.55% 37 3.02% 41 91.11%
30-34 4 0.59% 16 1.19% 20 4 0.77% 2 1.67% 6 30.00% 0.00% 14 1.14% 14 70.00%
35-39 2 0.30% 14 1.04% 16 1 0.19% 1 0.83% 2 12.50% 1 0.64% 13 1.06% 14 87.50%
40-49 1 0.15% 25 1.86% 26 0.00% 1 0.83% 1 3.85% 1 0.64% 24 1.96% 25 96.15%
50-64 0.00% 4 0.30% 4 0.00% 0.00% 0.00% 0.00% 4 0.33% 4 100.00%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Race/Ethnicity
American_Indian 2 0.30% 7 0.52% 9 2 0.39% 0.00% 2 22.22% 0.00% 7 0.57% 7 77.78%
Asian 5 0.74% 6 0.45% 11 5 0.97% 1 0.83% 6 54.55% 0.00% 5 0.41% 5 45.45%
Black 34 5.04% 159 11.81% 193 15 2.90% 3 2.50% 18 9.33% 19 12.10% 156 12.72% 175 90.67%
Hispanic 7 1.04% 36 2.67% 43 6 1.16% 2 1.67% 8 18.60% 1 0.64% 34 2.77% 35 81.40%
International 7 1.04% 5 0.37% 12 7 1.35% 2 1.67% 9 75.00% 0.00% 3 0.24% 3 25.00%
Unspecified_Race 40 5.93% 87 6.46% 127 29 5.61% 5 4.17% 34 26.77% 11 7.01% 82 6.69% 93 73.23%
White 579 85.91% 1,046 77.71% 1,625 453 87.62% 107 89.17% 560 34.46% 126 80.25% 939 76.59% 1,065 65.54%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Resident of Ohio
Yes 604 89.61% 1,207 89.67% 1,811 458 88.59% 102 85.00% 560 30.92% 146 92.99% 1,105 90.13% 1,251 69.08%
No 70 10.39% 139 10.33% 209 59 11.41% 18 15.00% 77 36.84% 11 7.01% 121 9.87% 132 63.16%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Commuter
Yes 524 77.74% 1,120 83.21% 1,644 398 76.98% 92 76.67% 490 29.81% 126 80.25% 1,028 83.85% 1,154 70.19%
No 150 22.26% 226 16.79% 376 119 23.02% 28 23.33% 147 39.10% 31 19.75% 198 16.15% 229 60.90%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Composite ACT Score Range
No ACT 70 10.39% 315 23.40% 385 51 9.86% 24 20.00% 75 19.48% 19 12.10% 291 23.74% 310 80.52%
2. 6-11 1 0.15% 5 0.37% 6 0.00% 0.00% 0.00% 1 0.64% 5 0.41% 6 100.00%
3. 12-17 102 15.13% 365 27.12% 467 58 11.22% 15 12.50% 73 15.63% 44 28.03% 350 28.55% 394 84.37%
4. 18-23 304 45.10% 511 37.96% 815 224 43.33% 52 43.33% 276 33.87% 80 50.96% 459 37.44% 539 66.13%
5. 24-29 173 25.67% 143 10.62% 316 160 30.95% 28 23.33% 188 59.49% 13 8.28% 115 9.38% 128 40.51%
6. 30-36 24 3.56% 7 0.52% 31 24 4.64% 1 0.83% 25 80.65% 0.00% 6 0.49% 6 19.35%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Table 1
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
High School Graduating GPA Ranges
Below_1.0 39 5.79% 122 9.06% 161 34 6.58% 6 5.00% 40 24.84% 5 3.18% 116 9.46% 121 75.16%
1.0-1.99 10 1.48% 175 13.00% 185 3 0.58% 3 2.50% 6 3.24% 7 4.46% 172 14.03% 179 96.76%
2.0-2.49 59 8.75% 275 20.43% 334 30 5.80% 8 6.67% 38 11.38% 29 18.47% 267 21.78% 296 88.62%
2.5-2.99 144 21.36% 329 24.44% 473 83 16.05% 25 20.83% 108 22.83% 61 38.85% 304 24.80% 365 77.17%
3.0-3.24 111 16.47% 189 14.04% 300 79 15.28% 25 20.83% 104 34.67% 32 20.38% 164 13.38% 196 65.33%
3.25-3.49 70 10.39% 89 6.61% 159 57 11.03% 17 14.17% 74 46.54% 13 8.28% 72 5.87% 85 53.46%
3.5-3.74 108 16.02% 78 5.79% 186 102 19.73% 21 17.50% 123 66.13% 6 3.82% 57 4.65% 63 33.87%
3.75_and_higher 131 19.44% 50 3.71% 181 128 24.76% 15 12.50% 143 79.01% 3 1.91% 35 2.85% 38 20.99%
GED_recipient 2 0.30% 36 2.67% 38 1 0.19% 0.00% 1 2.63% 1 0.64% 36 2.94% 37 97.37%
No GPA Information 0.00% 3 0.22% 3 0.00% 0.00% 0.00% 0.00% 3 0.24% 3 100.00%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Any Advanced Placement Credits
Yes 28 4.15% 13 0.97% 41 27 5.22% 6 5.00% 33 80.49% 1 0.64% 7 0.57% 8 19.51%
No 646 95.85% 1,333 99.03% 1,979 490 94.78% 114 95.00% 604 30.52% 156 99.36% 1,219 99.43% 1,375 69.48%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Academic Intention
Obtain_Associate_Degree_for_Job_Market 11 1.63% 35 2.60% 46 10 1.93% 2 1.67% 12 26.09% 1 0.64% 33 2.69% 34 73.91%
Obtain_Associate_Degree_for_Transfer 1 0.15% 1 0.07% 2 1 0.19% 0.00% 1 50.00% 0.00% 1 0.08% 1 50.00%
Obtain_Bachelors_Degree 313 46.44% 543 40.34% 856 235 45.45% 51 42.50% 286 33.41% 78 49.68% 492 40.13% 570 66.59%
Obtain_Undergraduate_Certificate 1 0.15% 1 0.07% 2 1 0.19% 0.00% 1 50.00% 0.00% 1 0.08% 1 50.00%
Personal_Interest 24 3.56% 87 6.46% 111 16 3.09% 8 6.67% 24 21.62% 8 5.10% 79 6.44% 87 78.38%
Selected_Courses_Train_New_Career 6 0.89% 20 1.49% 26 4 0.77% 1 0.83% 5 19.23% 2 1.27% 19 1.55% 21 80.77%
Selected_Courses_Upgrade_Skills 4 0.59% 10 0.74% 14 3 0.58% 0.00% 3 21.43% 1 0.64% 10 0.82% 11 78.57%
Transfer_Before_Degree 2 0.30% 12 0.89% 14 2 0.39% 0.00% 2 14.29% 0.00% 12 0.98% 12 85.71%
Unknown 312 46.29% 637 47.33% 949 245 47.39% 58 48.33% 303 31.93% 67 42.68% 579 47.23% 646 68.07%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Major Field of Study
Area, Ethnic, Cultural, Gender Studies 0.00% 1 0.07% 1 0.00% 0.00% 0.00% 0.00% 1 0.08% 1 100.00%
Biological and Biomedical Sciences 18 2.67% 58 4.31% 76 14 2.71% 4 3.33% 18 23.68% 4 2.55% 54 4.40% 58 76.32%
Business Management and Marketing 110 16.32% 204 15.16% 314 85 16.44% 22 18.33% 107 34.08% 25 15.92% 182 14.85% 207 65.92%
Communication and Journalism 8 1.19% 17 1.26% 25 3 0.58% 2 1.67% 5 20.00% 5 3.18% 15 1.22% 20 80.00%
Computer and Information Sciences 27 4.01% 90 6.69% 117 16 3.09% 6 5.00% 22 18.80% 11 7.01% 84 6.85% 95 81.20%
Education 143 21.22% 193 14.34% 336 111 21.47% 20 16.67% 131 38.99% 32 20.38% 173 14.11% 205 61.01%
Engineering 86 12.76% 88 6.54% 174 70 13.54% 9 7.50% 79 45.40% 16 10.19% 79 6.44% 95 54.60%
Engineering Technology 3 0.45% 24 1.78% 27 1 0.19% 1 0.83% 2 7.41% 2 1.27% 23 1.88% 25 92.59%
English Language and Literature 17 2.52% 37 2.75% 54 11 2.13% 3 2.50% 14 25.93% 6 3.82% 34 2.77% 40 74.07%
Family and Consumer Sciences 6 0.89% 18 1.34% 24 4 0.77% 2 1.67% 6 25.00% 2 1.27% 16 1.31% 18 75.00%
Foreign Languages and Literature 3 0.45% 3 0.22% 6 3 0.58% 2 1.67% 5 83.33% 0.00% 1 0.08% 1 16.67%
Health Professions and Clinical Services 28 4.15% 96 7.13% 124 21 4.06% 3 2.50% 24 19.35% 7 4.46% 93 7.59% 100 80.65%
Legal Professions and Studies 8 1.19% 18 1.34% 26 6 1.16% 3 2.50% 9 34.62% 2 1.27% 15 1.22% 17 65.38%
Leisure and Fitness Studies 22 3.26% 42 3.12% 64 15 2.90% 3 2.50% 18 28.13% 7 4.46% 39 3.18% 46 71.88%
Liberal Arts and General Studies 67 9.94% 154 11.44% 221 52 10.06% 13 10.83% 65 29.41% 15 9.55% 141 11.50% 156 70.59%
Mathematics and Statistics 3 0.45% 3 0.22% 6 3 0.58% 0.00% 3 50.00% 0.00% 3 0.24% 3 50.00%
Table 1
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
Natural Resources and Conservation 1 0.15% 1 0.07% 2 1 0.19% 0.00% 1 50.00% 0.00% 1 0.08% 1 50.00%
Philosophy and Religious Studies 2 0.30% 3 0.22% 5 1 0.19% 0.00% 1 20.00% 1 0.64% 3 0.24% 4 80.00%
Physical Sciences 17 2.52% 18 1.34% 35 16 3.09% 4 3.33% 20 57.14% 1 0.64% 14 1.14% 15 42.86%
Precision Production 1 0.15% 3 0.22% 4 1 0.19% 0.00% 1 25.00% 0.00% 3 0.24% 3 75.00%
Psychology 17 2.52% 54 4.01% 71 11 2.13% 10 8.33% 21 29.58% 6 3.82% 44 3.59% 50 70.42%
Public Administration and Social Service 5 0.74% 23 1.71% 28 3 0.58% 0.00% 3 10.71% 2 1.27% 23 1.88% 25 89.29%
Security and Protective Services 18 2.67% 74 5.50% 92 10 1.93% 0.00% 10 10.87% 8 5.10% 74 6.04% 82 89.13%
Social Sciences 13 1.93% 14 1.04% 27 13 2.51% 0.00% 13 48.15% 0.00% 14 1.14% 14 51.85%
Visual and Performing Arts 51 7.57% 110 8.17% 161 46 8.90% 13 10.83% 59 36.65% 5 3.18% 97 7.91% 102 63.35%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Student Marital Status
Life_Partner 0.00% 7 0.52% 7 0.00% 1 0.83% 1 14.29% 0.00% 6 0.49% 6 85.71%
Married 9 1.34% 35 2.60% 44 7 1.35% 3 2.50% 10 22.73% 2 1.27% 32 2.61% 34 77.27%
Single 665 98.66% 1,304 96.88% 1,969 510 98.65% 116 96.67% 626 31.79% 155 98.73% 1,188 96.90% 1,343 68.21%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Student Dependency Upon Parents
Dependent 499 74.04% 888 65.97% 1,387 400 77.37% 90 75.00% 490 35.33% 99 63.06% 798 65.09% 897 64.67%
Independent 23 3.41% 159 11.81% 182 10 1.93% 6 5.00% 16 8.79% 13 8.28% 153 12.48% 166 91.21%
Unspecified 149 22.11% 291 21.62% 440 107 20.70% 24 20.00% 131 29.77% 42 26.75% 267 21.78% 309 70.23%
Unspecified - X 3 0.45% 6 0.45% 9 0.00% 0.00% 0.00% 3 1.91% 6 0.49% 9 100.00%
Unspecified - Y 0.00% 2 0.15% 2 0.00% 0.00% 0.00% 0.00% 2 0.16% 2 100.00%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Cost of Attendance
1. No FAFSA on file 162 24.04% 343 25.48% 505 114 22.05% 28 23.33% 142 28.12% 48 30.57% 315 25.69% 363 71.88%
2. 9001-10000 289 42.88% 454 33.73% 743 233 45.07% 44 36.67% 277 37.28% 56 35.67% 410 33.44% 466 62.72%
3. 10001-12000 25 3.71% 50 3.71% 75 20 3.87% 9 7.50% 29 38.67% 5 3.18% 41 3.34% 46 61.33%
4. 12001-14000 152 22.55% 416 30.91% 568 115 22.24% 30 25.00% 145 25.53% 37 23.57% 386 31.48% 423 74.47%
5. 14001-16000 33 4.90% 56 4.16% 89 26 5.03% 9 7.50% 35 39.33% 7 4.46% 47 3.83% 54 60.67%
6. 16001-18000 13 1.93% 27 2.01% 40 9 1.74% 0.00% 9 22.50% 4 2.55% 27 2.20% 31 77.50%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Need Level Ranges - For those with a valid/complete FAFSA
1. -77999 to - 20000 14 2.73% 12 1.20% 26 12 2.98% 3 3.26% 15 57.69% 2 1.83% 9 0.99% 11 42.31%
2. -19999 to -10000 23 4.49% 23 2.30% 46 16 3.97% 2 2.17% 18 39.13% 7 6.42% 21 2.31% 28 60.87%
3. -9999 to -1 99 19.34% 127 12.67% 226 83 20.60% 12 13.04% 95 42.04% 16 14.68% 115 12.64% 131 57.96%
5. 1-2000 29 5.66% 43 4.29% 72 24 5.96% 10 10.87% 34 47.22% 0.00% 0.00% 0.00%
6. 2001-5000 64 12.50% 123 12.28% 187 52 12.90% 15 16.30% 67 35.83% 5 4.59% 33 3.63% 38 20.32%
7. 5001-8000 107 20.90% 151 15.07% 258 92 22.83% 14 15.22% 106 41.09% 12 11.01% 108 11.87% 120 46.51%
8. 8001-10000 104 20.31% 224 22.36% 328 78 19.35% 21 22.83% 99 30.18% 15 13.76% 137 15.05% 152 46.34%
9. 10001-12000 35 6.84% 69 6.89% 104 27 6.70% 8 8.70% 35 33.65% 26 23.85% 203 22.31% 229 220.19%
 10. 12001-14000 30 5.86% 204 20.36% 234 17 4.22% 7 7.61% 24 10.26% 8 7.34% 61 6.70% 69 29.49%
 11. Over 14001 7 1.37% 26 2.59% 33 2 0.50% 0.00% 2 6.06% 13 11.93% 197 21.65% 210 636.36%
Grand Total 512 100.00% 1,002 100.00% 1,514 403 100.00% 92 100.00% 495 32.69% 5 4.59% 26 2.86% 31 2.05%
Table 1
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
9-Month Expected Family Contribution - For those with a valid/complete FAFSA
1. No_Family_Contribution 51 9.96% 259 25.85% 310 32 7.94% 11 11.96% 43 13.87% 19 17.43% 248 27.25% 267 86.13%
2. 1-1000 27 5.27% 100 9.98% 127 15 3.72% 5 5.43% 20 15.75% 12 11.01% 95 10.44% 107 84.25%
3. 1001-3000 94 18.36% 142 14.17% 236 71 17.62% 13 14.13% 84 35.59% 23 21.10% 129 14.18% 152 64.41%
4. 3001-5000 86 16.80% 131 13.07% 217 76 18.86% 14 15.22% 90 41.47% 10 9.17% 117 12.86% 127 58.53%
5. 5001-7000 53 10.35% 101 10.08% 154 45 11.17% 14 15.22% 59 38.31% 8 7.34% 87 9.56% 95 61.69%
6. 7001-9000 36 7.03% 58 5.79% 94 29 7.20% 10 10.87% 39 41.49% 7 6.42% 48 5.27% 55 58.51%
7. 9001-11000 23 4.49% 52 5.19% 75 21 5.21% 6 6.52% 27 36.00% 2 1.83% 46 5.05% 48 64.00%
8. 11001-13000 34 6.64% 35 3.49% 69 25 6.20% 2 2.17% 27 39.13% 9 8.26% 33 3.63% 42 60.87%
9. 13001-20000 69 13.48% 82 8.18% 151 59 14.64% 12 13.04% 71 47.02% 10 9.17% 70 7.69% 80 52.98%
10. 20001-40000 34 6.64% 39 3.89% 73 26 6.45% 3 3.26% 29 39.73% 8 7.34% 36 3.96% 44 60.27%
11. Over 40001 5 0.98% 3 0.30% 8 4 0.99% 2 2.17% 6 75.00% 1 0.92% 1 0.11% 2 25.00%
Grand Total 512 100.00% 1,002 100.00% 1,514 403 100.00% 92 100.00% 495 32.69% 109 100.00% 910 100.00% 1,019 67.31%
Any Aid
Yes 670 99.41% 1,325 98.44% 1,995 513 99.23% 118 98.33% 631 31.63% 157 100.00% 1,207 98.45% 1,364 68.37%
No 4 0.59% 21 1.56% 25 4 0.77% 2 1.67% 6 24.00% 0.00% 19 1.55% 19 76.00%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Federal Aid (Excluding Student Loans) Ranges - For those with a valid/complete FAFSA
1. No_Aid 319 62.30% 473 47.21% 792 264 65.51% 63 68.48% 327 41.29% 55 50.46% 410 45.05% 465 58.71%
2. 1-199 0.00% 3 0.30% 3 0.00% 0.00% 0.00% 0.00% 3 0.33% 3 100.00%
3. 200-499 28 5.47% 46 4.59% 74 26 6.45% 1 1.09% 27 36.49% 2 1.83% 45 4.95% 47 63.51%
4. 500-749 21 4.10% 34 3.39% 55 17 4.22% 1 1.09% 18 32.73% 4 3.67% 33 3.63% 37 67.27%
5. 750-999 29 5.66% 42 4.19% 71 21 5.21% 3 3.26% 24 33.80% 8 7.34% 39 4.29% 47 66.20%
6. 1000-1249 22 4.30% 46 4.59% 68 15 3.72% 5 5.43% 20 29.41% 7 6.42% 41 4.51% 48 70.59%
7. 1250-1499 24 4.69% 37 3.69% 61 17 4.22% 1 1.09% 18 29.51% 7 6.42% 36 3.96% 43 70.49%
8. 1500-1749 15 2.93% 38 3.79% 53 7 1.74% 4 4.35% 11 20.75% 8 7.34% 34 3.74% 42 79.25%
9. 1750-1999 45 8.79% 236 23.55% 281 31 7.69% 13 14.13% 44 15.66% 14 12.84% 223 24.51% 237 84.34%
10. 2000-2249 6 1.17% 20 2.00% 26 4 0.99% 1 1.09% 5 19.23% 2 1.83% 19 2.09% 21 80.77%
11. 2250-2499 3 0.59% 27 2.69% 30 1 0.25% 0.00% 1 3.33% 2 1.83% 27 2.97% 29 96.67%
12. 2500-2749 0.00% 0.00% 0.00% 0.00% #DIV/0! 0.00% 0.00% #DIV/0!
Grand Total 512 100.00% 1,002 100.00% 1,514 403 100.00% 92 100.00% 495 32.69% 109 100.00% 910 100.00% 1,019 67.31%
State Aid Ranges - For those with a valid/complete FAFSA
1. No_State_Aid 355 69.34% 553 55.19% 908 294 72.95% 62 67.39% 356 39.21% 61 55.96% 491 53.96% 552 60.79%
2. 1-500 87 16.99% 216 21.56% 303 60 14.89% 18 19.57% 78 25.74% 27 24.77% 198 21.76% 225 74.26%
3. 501-1000 36 7.03% 127 12.67% 163 25 6.20% 9 9.78% 34 20.86% 11 10.09% 118 12.97% 129 79.14%
4. 1001-1500 33 6.45% 99 9.88% 132 24 5.96% 3 3.26% 27 20.45% 9 8.26% 96 10.55% 105 79.55%
6. 2001-2500 1 0.20% 6 0.60% 7 0.00% 0.00% 0.00% 1 0.92% 6 0.66% 7 100.00%
7. Over 2500 0.00% 1 0.10% 1 0.00% 0.00% 0.00% 0.00% 1 0.11% 1 100.00%
Grand Total 512 100.00% 1,002 100.00% 1,514 403 100.00% 92 100.00% 495 32.69% 109 100.00% 910 100.00% 1,019 67.31%
Table 1
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
Federal Work Study Aid Ranges - For those with a valid/complete FAFSA
1. No_Work_Study 498 97.27% 978 97.60% 1,476 393 97.52% 89 96.74% 482 32.66% 105 96.33% 889 97.69% 994 67.34%
2. 1-250 1 0.20% 7 0.70% 8 1 0.25% 0.00% 1 12.50% 0.00% 7 0.77% 7 87.50%
3. 251-500 3 0.59% 10 1.00% 13 1 0.25% 2 2.17% 3 23.08% 2 1.83% 8 0.88% 10 76.92%
4. 501-750 5 0.98% 0.00% 5 3 0.74% 0.00% 3 60.00% 2 1.83% 0.00% 2 40.00%
5. 751-1000 3 0.59% 3 0.30% 6 3 0.74% 1 1.09% 4 66.67% 0.00% 2 0.22% 2 33.33%
6. 1001-1250 1 0.20% 1 0.10% 2 1 0.25% 0.00% 1 50.00% 0.00% 1 0.11% 1 50.00%
7. 1251-1500 0.00% 2 0.20% 2 0.00% 0.00% 0.00% 0.00% 2 0.22% 2 100.00%
8. 1501-1750 1 0.20% 1 0.10% 2 1 0.25% 0.00% 1 50.00% 0.00% 1 0.11% 1 50.00%
Grand Total 512 100.00% 1,002 100.00% 1,514 403 100.00% 92 100.00% 495 32.69% 109 100.00% 910 100.00% 1,019 67.31%
Institutional Aid Ranges
1. No_Institutional_Aid 280 41.54% 958 71.17% 1,238 170 32.88% 53 44.17% 223 18.01% 110 70.06% 905 73.82% 1,015 81.99%
2. 1-500 174 25.82% 244 18.13% 418 148 28.63% 38 31.67% 186 44.50% 26 16.56% 206 16.80% 232 55.50%
3. 501-1000 80 11.87% 63 4.68% 143 72 13.93% 13 10.83% 85 59.44% 8 5.10% 50 4.08% 58 40.56%
4. 1001-1500 57 8.46% 35 2.60% 92 54 10.44% 4 3.33% 58 63.04% 3 1.91% 31 2.53% 34 36.96%
5. 1501-2000 17 2.52% 13 0.97% 30 16 3.09% 3 2.50% 19 63.33% 1 0.64% 10 0.82% 11 36.67%
6. 2001-3000 15 2.23% 14 1.04% 29 11 2.13% 4 3.33% 15 51.72% 4 2.55% 10 0.82% 14 48.28%
7. 3001-4000 7 1.04% 6 0.45% 13 5 0.97% 1 0.83% 6 46.15% 2 1.27% 5 0.41% 7 53.85%
8. 4001-5000 7 1.04% 2 0.15% 9 4 0.77% 0.00% 4 44.44% 3 1.91% 2 0.16% 5 55.56%
9. 5001-6000 26 3.86% 5 0.37% 31 26 5.03% 2 1.67% 28 90.32% 0.00% 3 0.24% 3 9.68%
 10. 6001-7000 5 0.74% 3 0.22% 8 5 0.97% 1 0.83% 6 75.00% 0.00% 2 0.16% 2 25.00%
 11. 7001-8000 6 0.89% 3 0.22% 9 6 1.16% 1 0.83% 7 77.78% 0.00% 2 0.16% 2 22.22%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Other Third Party Aid Ranges
1. No_3rd_Party_Aid 18 2.67% 115 8.54% 133 10 1.93% 4 3.33% 14 10.53% 8 5.10% 111 9.05% 119 89.47%
2. 1-100 262 38.87% 768 57.06% 1,030 178 34.43% 48 40.00% 226 21.94% 84 53.50% 720 58.73% 804 78.06%
3. 101-500 23 3.41% 44 3.27% 67 22 4.26% 7 5.83% 29 43.28% 1 0.64% 37 3.02% 38 56.72%
4. 501-600 239 35.46% 273 20.28% 512 196 37.91% 36 30.00% 232 45.31% 43 27.39% 237 19.33% 280 54.69%
5. 601-1000 31 4.60% 27 2.01% 58 28 5.42% 8 6.67% 36 62.07% 3 1.91% 19 1.55% 22 37.93%
6. 1001-1500 44 6.53% 53 3.94% 97 37 7.16% 9 7.50% 46 47.42% 7 4.46% 44 3.59% 51 52.58%
7. 1501-2500 43 6.38% 47 3.49% 90 36 6.96% 6 5.00% 42 46.67% 7 4.46% 41 3.34% 48 53.33%
8. 2501-5000 12 1.78% 18 1.34% 30 9 1.74% 2 1.67% 11 36.67% 3 1.91% 16 1.31% 19 63.33%
9. 5001-8000 2 0.30% 1 0.07% 3 1 0.19% 0.00% 1 33.33% 1 0.64% 1 0.08% 2 66.67%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Table 1
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
Student Loan Ranges
1. No_Student_Loan 435 64.54% 718 53.34% 1,153 346 66.92% 73 60.83% 419 36.34% 89 56.69% 645 52.61% 734 63.66%
2. 1-1000 16 2.37% 45 3.34% 61 11 2.13% 3 2.50% 14 22.95% 5 3.18% 42 3.43% 47 77.05%
3. 1001-2000 145 21.51% 308 22.88% 453 112 21.66% 27 22.50% 139 30.68% 33 21.02% 281 22.92% 314 69.32%
4. 2001-3000 27 4.01% 82 6.09% 109 16 3.09% 4 3.33% 20 18.35% 11 7.01% 78 6.36% 89 81.65%
5. 3001-4000 23 3.41% 122 9.06% 145 13 2.51% 6 5.00% 19 13.10% 10 6.37% 116 9.46% 126 86.90%
6. 4001-5000 12 1.78% 25 1.86% 37 7 1.35% 0.00% 7 18.92% 5 3.18% 25 2.04% 30 81.08%
7. 5001-6000 11 1.63% 31 2.30% 42 8 1.55% 5 4.17% 13 30.95% 3 1.91% 26 2.12% 29 69.05%
8. 6001-7000 4 0.59% 10 0.74% 14 3 0.58% 2 1.67% 5 35.71% 1 0.64% 8 0.65% 9 64.29%
9. 7001-8000 0.00% 3 0.22% 3 0.00% 0.00% 0.00% 0.00% 3 0.24% 3 100.00%
10. 8001-8500 1 0.15% 2 0.15% 3 1 0.19% 0.00% 1 33.33% 0.00% 2 0.16% 2 66.67%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
First Term Credit Hour Load
Full-Time 660 97.92% 1,240 92.12% 1,900 514 99.42% 120 100.00% 634 33.37% 146 92.99% 1,120 91.35% 1,266 66.63%
Part-Time 14 2.08% 106 7.88% 120 3 0.58% 0.00% 3 2.50% 11 7.01% 106 8.65% 117 97.50%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
First Term Attempted Credit Hours
1. 0 0.00% 91 6.76% 91 0.00% 0.00% 0.00% 0.00% 91 7.42% 91 100.00%
2. 1-5 2 0.30% 82 6.09% 84 0.00% 0.00% 0.00% 2 1.27% 82 6.69% 84 100.00%
3. 6-10 32 4.75% 315 23.40% 347 4 0.77% 0.00% 4 1.15% 28 17.83% 315 25.69% 343 98.85%
4. 11-15 487 72.26% 749 55.65% 1,236 371 71.76% 78 65.00% 449 36.33% 116 73.89% 671 54.73% 787 63.67%
5. 16-19 153 22.70% 109 8.10% 262 142 27.47% 42 35.00% 184 70.23% 11 7.01% 67 5.46% 78 29.77%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
First Term Total Credit Hours Earned
0 0.00% 256 19.02% 256 0.00% 0.00% 0.00% 0.00% 256 20.88% 256 100.00%
1.00-6.00 10 1.48% 228 16.94% 238 2 0.39% 0.00% 2 0.84% 8 5.10% 228 18.60% 236 99.16%
7.00-11.00 44 6.53% 247 18.35% 291 4 0.77% 2 1.67% 6 2.06% 40 25.48% 245 19.98% 285 97.94%
12.00-16.00 525 77.89% 543 40.34% 1,068 427 82.59% 91 75.83% 518 48.50% 98 62.42% 452 36.87% 550 51.50%
17.00-21.00 72 10.68% 57 4.23% 129 65 12.57% 22 18.33% 87 67.44% 7 4.46% 35 2.85% 42 32.56%
22.00-26.00 8 1.19% 2 0.15% 10 7 1.35% 1 0.83% 8 80.00% 1 0.64% 1 0.08% 2 20.00%
27.00-31.00 8 1.19% 3 0.22% 11 7 1.35% 3 2.50% 10 90.91% 1 0.64% 0.00% 1 9.09%
32.00-over 7 1.04% 10 0.74% 17 5 0.97% 1 0.83% 6 35.29% 2 1.27% 9 0.73% 11 64.71%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
First Term Cumulative Quality Points
1. 0 0.00% 273 20.28% 273 0.00% 0.00% 0.00% 0.00% 273 22.27% 273 100.00%
2. 1-12 5 0.74% 190 14.12% 195 0.00% 0.00% 0.00% 5 3.18% 190 15.50% 195 100.00%
3. 13-24 50 7.42% 253 18.80% 303 3 0.58% 0.00% 3 0.99% 47 29.94% 253 20.64% 300 99.01%
4. 25-36 107 15.88% 279 20.73% 386 30 5.80% 3 2.50% 33 8.55% 77 49.04% 276 22.51% 353 91.45%
5. 37-48 240 35.61% 231 17.16% 471 216 41.78% 44 36.67% 260 55.20% 24 15.29% 187 15.25% 211 44.80%
6. 49-60 216 32.05% 105 7.80% 321 212 41.01% 69 57.50% 281 87.54% 4 2.55% 36 2.94% 40 12.46%
7. 61-76 56 8.31% 15 1.11% 71 56 10.83% 4 3.33% 60 84.51% 0.00% 11 0.90% 11 15.49%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Table 1
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
First Term GPA Ranges
1. Below 1.0 0.00% 356 26.45% 356 0.00% 0.00% 0.00% 0.00% 356 29.04% 356 100.00%
2. 1.00-1.99 37 5.49% 236 17.53% 273 2 0.39% 0.00% 2 0.73% 35 22.29% 236 19.25% 271 99.27%
3. 2.00-2.49 66 9.79% 225 16.72% 291 18 3.48% 3 2.50% 21 7.22% 48 30.57% 222 18.11% 270 92.78%
4. 2.50-2.99 121 17.95% 189 14.04% 310 90 17.41% 16 13.33% 106 34.19% 31 19.75% 173 14.11% 204 65.81%
5. 3.00-3.24 95 14.09% 137 10.18% 232 73 14.12% 28 23.33% 101 43.53% 22 14.01% 109 8.89% 131 56.47%
6. 3.25-3.49 107 15.88% 78 5.79% 185 96 18.57% 27 22.50% 123 66.49% 11 7.01% 51 4.16% 62 33.51%
7. 3.50-3.74 88 13.06% 58 4.31% 146 83 16.05% 22 18.33% 105 71.92% 5 3.18% 36 2.94% 41 28.08%
8. 3.75 and higher 160 23.74% 67 4.98% 227 155 29.98% 24 20.00% 179 78.85% 5 3.18% 43 3.51% 48 21.15%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Any Remediation
Yes 311 46.14% 804 59.73% 1,115 203 39.26% 53 44.17% 256 22.96% 108 68.79% 751 61.26% 859 77.04%
No 363 53.86% 542 40.27% 905 314 60.74% 67 55.83% 381 42.10% 49 31.21% 475 38.74% 524 57.90%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Remedial English
Did_not_take 455 67.51% 820 60.92% 1,275 382 73.89% 88 73.33% 470 36.86% 73 46.50% 732 59.71% 805 63.14%
Failed 0.00% 103 7.65% 103 0.00% 1 0.83% 1 0.97% 0.00% 102 8.32% 102 99.03%
Passed 219 32.49% 423 31.43% 642 135 26.11% 31 25.83% 166 25.86% 84 53.50% 392 31.97% 476 74.14%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Remedial Mathematics
Did_not_take 480 71.22% 791 58.77% 1,271 393 76.02% 86 71.67% 479 37.69% 87 55.41% 705 57.50% 792 62.31%
Failed 5 0.74% 213 15.82% 218 0.00% 1 0.83% 1 0.46% 5 3.18% 212 17.29% 217 99.54%
Passed 189 28.04% 342 25.41% 531 124 23.98% 33 27.50% 157 29.57% 65 41.40% 309 25.20% 374 70.43%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Reading & Study Skills Course Work
Did_not_take 557 82.64% 1,031 76.60% 1,588 449 86.85% 102 85.00% 551 34.70% 108 68.79% 929 75.77% 1,037 65.30%
Failed 0.00% 51 3.79% 51 0.00% 0.00% 0.00% 0.00% 51 4.16% 51 100.00%
Passed 117 17.36% 264 19.61% 381 68 13.15% 18 15.00% 86 22.57% 49 31.21% 246 20.07% 295 77.43%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Center for Student Profess # of Visits
None 499 74.04% 1,213 90.12% 1,712 380 73.50% 94 78.33% 474 27.69% 119 75.80% 1,119 91.27% 1,238 72.31%
1. 1-5 136 20.18% 110 8.17% 246 106 20.50% 22 18.33% 128 52.03% 30 19.11% 88 7.18% 118 47.97%
2. 6-10 21 3.12% 10 0.74% 31 17 3.29% 2 1.67% 19 61.29% 4 2.55% 8 0.65% 12 38.71%
3. 11-15 14 2.08% 8 0.59% 22 10 1.93% 2 1.67% 12 54.55% 4 2.55% 6 0.49% 10 45.45%
4. 16-20 4 0.59% 2 0.15% 6 4 0.77% 0.00% 4 66.67% 0.00% 2 0.16% 2 33.33%
5. 21+ 0.00% 3 0.22% 3 0.00% 0.00% 0.00% 0.00% 3 0.24% 3 100.00%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Earned an Associate Degree
Yes 25 3.71% 48 3.57% 73 10 1.93% 1 0.83% 11 15.07% 15 9.55% 47 3.83% 62 84.93%
No 649 96.29% 1,298 96.43% 1,947 507 98.07% 119 99.17% 626 32.15% 142 90.45% 1,179 96.17% 1,321 67.85%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Table 1
Appendix B
2001 First-Time J48 Predicted to J48 Predicted NOT to
Undergraduate Cohort Earn a Bachelor Degree Earned a Bachelor Degree
Actually Earned Actually Earned Actually Earned
Bachelor Degree Bachelor Degree Part of Bachelor Degree Part of
Yes No Cohort Yes No Cohort Yes No Cohort
# % # % # # % # % # % # % # % # %
Continued to Following Spring Term
Yes 669 99.26% 1,046 77.71% 1,715 514 99.42% 117 97.50% 631 36.79% 155 98.73% 929 75.77% 1,084 63.21%
No 5 0.74% 300 22.29% 305 3 0.58% 3 2.50% 6 1.97% 2 1.27% 297 24.23% 299 98.03%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Consecutively Enrolled (Fall, Spring, Fall)
Yes 654 97.03% 705 52.38% 1,359 507 98.07% 117 97.50% 624 45.92% 147 93.63% 588 47.96% 735 54.08%
No 20 2.97% 641 47.62% 661 10 1.93% 3 2.50% 13 1.97% 10 6.37% 638 52.04% 648 98.03%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Returned Next Fall
Yes 656 97.33% 733 54.46% 1,389 509 98.45% 120 100.00% 629 45.28% 147 93.63% 613 50.00% 760 54.72%
No 18 2.67% 613 45.54% 631 8 1.55% 0.00% 8 1.27% 10 6.37% 613 50.00% 623 98.73%
Grand Total 674 100.00% 1,346 100.00% 2,020 517 100.00% 120 100.00% 637 31.53% 157 100.00% 1,226 100.00% 1,383 68.47%
Table 1