An optimized random forest for multi class classification to classify the students using machine learning approaches

Abstract

In recent years, Multi-class classification in Educational Data Mining (EDM) has been continued to be a focusing research to deal the issues in the imbalanced data set and less number of unifying classification algorithms. Machine-learning algorithms take prominent part in multi-class classification. This research studyintroduced an Optimized Random Forest for Multi-Class Classification (ORFMCC) to classify the students based on the higher education programs enrolment. The base classifier Random Forest is optimized by hyper-parameter tuning and feature selection processes. The Optimized RFMCC is developed in Python 3.3 using Spider IDE 4.1.5. The optimum parameter extracted by tuning and relevant features significantly improved classification accuracy. The classification performance of Optimized Random Forest for Multi-Class Classifieris compared with RF, NB, DT, LR, and KNN. The experiments result of Accuracy, F1-Score, precision and Recall revealed that the ORFMCC outperformed in Multi-Class classification compared to the other five Classifiers.