Analysis Comparison of Data Mining Algorithm for Prediction Student Graduation Target

Abstract

The main objective of a higher education institution is to provide quality education for its students. The most important indicator to measure the quality of higher education performance is the percentage of student graduation on time. However, not all student can successfully have completed their studies during the four years of normal study period where it became problems for academic planners. So, it can affect to the study program accreditation assessment. In this study, C4.5 algorithms and fuzzy AHP are used to predict the number of students graduating on time. An analysis has conducted on how students can graduate on time and plan strategies for groups of students who are likely not to graduate on time. Furthermore, a comparative analysis of the algorithms that have been implemented which will provide more precise and accurate results. Data processing is carried out using the Rapidminer application. The results of the student graduation target analysis were found that the main factor of graduation on time using FAHP was the number of repeating courses (do not pass courses), while in C4.5 it was caused by GPA level 2. Both algorithms had a good level of accuracy, where FAHP and C4.5 were 100% and 82.24% respectively. This research can be used as a reference basis for supporting academic planners in making the right decisions for student groups produced so that all students can graduate on time.

Students Graduation Prediction Based on Academic Data Record Using the Decision Tree Algorithm C4.5 Method

“…After that, enter 2017 data that has been adjusted with a total of 368 records as the implementation of the model. Determination of independent attributes is based on references [8] and selected three attributes, namely; TAK, Parents' Income and School Origin, then use several other attributes because they are in accordance with the research objectives. Based on the influence of the selection of student specialization on the prediction of graduation and also based on the student's academic value, the following is an explanation of the attributes used as predictors in the input data : Income, containing information on the income of parents (guardians) of students based on Gross Domestic Product (GDP) per capita, Parent's income is also used as a graduation prediction factor in the study [18].…”

“…Well-known algorithms for predicting on-time graduation rates are ID3, CART, Naïve Bayes, fuzzy AHP (FAHP) and C4.5 algorithms [8]. The C4.5 algorithm has the highest performance value and level of accuracy compared to ID3 and CART in classifying [9] and against Naïve Bayes the C4.5 algorithm also has higher accuracy in determining the time of graduation on time with quite a lot of variables [10].…”

Educational Anomaly Analytics: Features, Methods, and Challenges

“…Likely, Yu et al (2018b) also apply logistic regression model to analyse the importance of high school academic preparation and postsecondary academic support services for the prediction of college completion among students with learning disabilities. Andreswari et al (2019) use C4.5 algorithms, a type of decision tree algorithm, to explore the relationship between student graduation, and academic performance, and family factors like parents' jobs and income. Purnamasari et al (2019) also use C4.5 algorithms to explore how academic performance can impact the final graduation time of students.…”

“…Ojha et al (2017) and Tampakas et al (2018) use demographic features and historical academic features to predict students' graduation time. Andreswari et al (2019) introduce more detailed background information. In addition to demographic features and historical academic features, they also add parents' jobs and income for graduation time prediction, and the results demonstrate the effectiveness of these features.…”