Academic analytics system for the College of Computer Studies of Cebu Institute of Technology - University / by Cherry Lyn C. Sta. Romana

Thesis (DIT) -- Cebu Institute of Technology - University, College of Computer Studies, March 2018

Academic Analytics is being increasingly used to analyze data stored in academic management systems to come up with policies that will increase graduation rate and increase retention rate. Cebu Institute of Technology – University is a private university located in Cebu City, Philippines. The school employs a liberal admission policy but supports students using various mechanisms like bridging, peer-learning and learning enhancement programs. Despite these student support programs, attrition rate remains high and graduation rate remains low. There is thus a need to craft data- driven policies in order to have a more targeted and systematic program for student admission, support and retention.
The data used in the study are the student records of BS Computer Science (BSCS) and BS Information Technology (BSIT) students of the College of Computer Studies, one of the colleges of CIT-U with high attrition rate. Date of students who entered the University from 2010 to 2013 were extracted. Based on the data, graduation rate is only 34% while retention rate going into the 3rd year level is only at 50%. In order to address these concerns, an Academic Analytics Systems was developed. A descriptive academic analytics systems was first created that allows a more cohesive and effective manner of presenting date that can be used for better data exploration. The system includes enrollment status, passing / failing rate, student performance, demographics and attrition/ retention. The system presents aggregated reports visually and interactively. Users can click the components of a graph to get more detailed information. The graphical representations allow for patterns and trends to be uncovered.
After a thorough exploration of the reports, predictive analytics using R was applied to the data. Logistics regression was used to come- up with a model for predicts successful graduation based on Math (3 courses) and English grades (2 courses). The model when tested yielded an accuracy rate of 76%. Another version was created but with major as one of the predictors After testing, the model yielded a slightly higher accuracy rate of 77%. Regression tree was also created to which shows that English and Math are important predictors of graduation.
Grades from the first year of College which includes grades in math, computer courses and major (BSIT or BSCS) were also used to predict attrition. The logistics regression model yielded an accuracy rate of 93%. A regression tree was also created which shows that BSIT students persist after the first years despite low grades. However, for Computer Science students, only with 3.5 grades in Programing 2 (CCS122) are retained in the program.
The last model that was created is for determining a university predicted grade using linear regression model when tested obtained a root mean squared error of 0.21 and adjusted r2 of 0.81. The model also shows that Statistics and English are important predictors. Based on these models, policy recommendations were drafted for student admission, support and retention.