Machine Learning Algorithms and Ensemble Technique to Improve Prediction of Students Performance

Abstract

Measuring Student’s performance is necessary for the mordent society. Applications of Machine learning algorithms increased the growth in various fields like disease prediction, student’s performance prediction, and crop productions prediction and in various other fields. The main aim of this study is to improve the prediction of students’ performance using various machine learning algorithms and ensemble technique to get better accuracy over individual machine learning algorithms. We have used the student’s dataset, which consists of 1000 instances and 22 attributes for evaluating the performance of students. In this paper we have applied four machine learning algorithms Decision Tree (DT), Naïve Bayesian (NB), K-Nearest Neighbors (KNN) and Extra Tree (ET) and then we have developed a model to combine the results of each individual base learner using Bagging and Boosting ensemble methods. The results obtained using bagging and boosting ensemble techniques were compared to select the best model. The results of all machine learning algorithms and ensemble techniques are tested with various factors like accuracy, sensitivity, specificity and f1-score. After comparison of results we find that bagging is the best method which gives the better result as compared to bagging ensemble techniques. The developed model can be applied on the admission seeking students to identify the perdition of their performance in the selected course, which can be beneficial for both the students and Institution.