Ant Colony Optimization Based Feature Subset Selection with Logistic Regression Classification Model for Education Data Mining

Abstract

In recent times, Educational data mining (EDM) becomes a more popular research area and data mining concepts are utilized in educational sector to extract meaningful details on the characteristics of the students in the learning task. In EDM, feature selection task is essential to generate a subset of candidate parameters. Since the feature selection (FS) process plays a main role in the classifier outcome, it is needed to analyze the efficiency of student assessment model with FS approaches. In this view, this study presents an effective FS based classification model for EDM. Here, ant colony optimization (ACO) algorithm is used for the selection of the most appropriate subset features with minimal cardinality to achieve better classifier performance. Then, logistic regression (LR) Classification technique is used to classify the educational data. The proposed ACO-LR model is validated using a benchmark Student Performance Data Set. The simulation outcome showed better classification of the presented model over the compared methods under diverse aspects. The experimental results showed that the ACO-LR model has offered a maximum precision of 97.99%, recall of 96.07%, accuracy of 94.91%, F-measure of 97.02% and kappa value of 79.57% respectively.