Abstract
Recently, education sectors have the most attraction from people all around the world and this make it more valuable for those wanting to invest in this sector and earn income. Students are the largest stakeholders in this area and they need more attention from the educational side. All universities are trying to improve quality for achieving their students’ satisfaction. Sometimes universities are using their old data related to students to analyze and make better decision for their future aspects. These analyses can be done by EDM (Educational Data Mining) a subset of ML (Machine Learning) that can discover very large datasets for producing valuable results. In this study we analyze the data of Computer Engineering students to predict their academic performances in three different aspects: predict final grades, study duration, and next term course grade. For this purpose, we have examined different ML algorithms and different available features to find the best ML algorithm and the most significant factors for predicting students’ academic performance. Our results show that SVM (Support Vector Machine) and DT (Decision Tree) are the two best ML algorithms and also, we have been determined that only course grades are the most valuable factors in prediction.
Presenters
Cansu Cigdem EkinAssociate Professor, Computer Engineering, Atilim University, Ankara, Turkey
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
KEYWORDS
Machine Learning, Data Mining, Educational Data Mining, Students’ Success Rate
