Predicting the Learning Behavior of Students Based on Personality Trait and Comparing the Classification Algorithms

Educational Data Mining (EDM) is a recent trend where data mining methods are experiments for the improvement of student performance in academics. This work describes the method of measuring the students learning behavior based on PTT (personality trait test). An experiment is attempted to improve the accuracy for predicting which student's Personality Trait uses all the available attributes. The student behavior was measured into five personality trait variables using PTQ (Personality trait Questionnaire). These variables are extroversion, agreeableness, consciousness, neuroticism, openness to experience. The data mining technique of attribute selection is applied for selecting the best attributes. Attribute selection is done by using WEKA tool. Data is rebalanced using cost responsive classification. The classification algorithms naive bayes, J48, Random forest, AD3, and Random tree are grouped with analysis and was experimented using Weka tool. The resultant analysis is compared with learning behavior of the student data based on the dataset PTQ and Classification methods.