A Multi-level Classification Model Pertaining to the Student’s Academic Performance Prediction

The students’ performance monitoring and evaluation is an essential activity of an education system to keep track of the success and failure records of the students. The objective of this research is to provide the best classification model to predict the students’ academic performance. In this paper we propose a Multilevel Classification Model (MLCM) based on Decision Tree Algorithm for the predictions of the academic performance of the undergraduate engineering students. The multi-level classification model consists of two levels. In level one, the four classification models namely Decision Tree (J48), Lazy Learner (IBK), Neural Network (MLP) and Naïve Bayes Tree (NBT) were constructed, evaluated and compared. The decision tree classifier was selected for the model construction in this step. In level 2, the overall accuracy of the classification model as well as the accuracy of individual class was enhanced by eliminating the outliers from the original dataset and by constructing Multilevel Classification Model (MLCM) using filtered dataset.