Predicting Students’ Drop-Out Rate Using Machine Learning Models: A Comparative Study

One of the major challenges facing Private Higher Educational Institutes in India is to reduce drop-out rate of first year students. The problem has got exacerbated post-covid specially for Engineering and management discipline. This research is an empirical study on efficiency and accuracy of various Machine Learning (ML) Prediction algorithms to predict the drop-out rate of students based on dataset available on predictors such as Family size, Study time, Time spent on extra- curricular activities, Time spent on Internet, Health, Absenteeism etc. A comparison of the performance of the ML models based on ‘Accuracy’ and ‘F1 score’ (to cater for variations in costs of false positives and false negatives) has been made to identify the best algorithm for given problem. This would help HEIs to identify potential drop-out students beforehand and take course correction measures thus improving retention. The study is conducted for B.Tech First year students with a sample size of 395 students using Logistic regression and K-NN algorithms. This preliminary work could be extended using other ML models such as; Support Vector Machine (SVM), Naïve Bayes, Decision Tree etc. or a combination of in an ensemble fashion in future.