Feature Correlation with Student Education Performance

Abstract

The 21st century has seen the advent of the internet as well as the spread of increasingly powerful computer technologies. One of these new technologies is Artificial Intelligence and Machine Learning. These computer models assist in pattern recognition, task performance as well as prediction. One place where this technology can be used is Educational Data Mining. This study used these ML technologies on the Student Performance Dataset to see what features are correlated with high student academic performance. This study also utilized Feature Engineering to derive features that represent the interactions of different features from the original dataset in order to conduct further analysis.This study found that multiple different features such as parent relationship status, travel time between home and school, among others, had a positive correlation with student academic performance. Features such as past failures and increasing frequency of hanging out with friends after school was correlated with negative student academic performance. However, results with the ML models as well as Feature Engineering were inconclusive due to the results not having a high enough accuracy to merit analysis.