Predicting Students Academic Performance using an Improved Random Forest Classifier
Sujith Jayaprakash, Sangeetha Krishnan, V. Jaiganesh
Abstract
Increasing demand in the education sector has paved the way to many research works which highly focuses on the student's academic performance and their behaviour analysis. Machine learning algorithms are applied in Education Data mining to find meaningful patterns and insights from the Educational datasets. Mostly, it is used to foretell the performance of students; to classify the features that impact the performance of students and also to cluster the students based on their performance. Introducing an interim mechanism to analyse and predict the student performance at the early stage help the institution and the students to take immediate action and improve the performance. In this research work, we have discussed the two most important factors – Factors that impact the scholastic achievement of students and aid in predicting the students at risk. This study also proposes a technique named improved random forest classifier. This technique aims to produce a higher accuracy rate in classification and prediction in comparison with the other algorithms such as Naive Bayes, Bagging, Boosting and Random Forest.