Predicting Student Academic Performance Using Machine Learning Techniques

In schools student performance is still judged mainly by their final results, which does not always help in finding students who may be struggling at an early stage. Student performance depends on things like how much time they spend studying how often they attend classes what grades they got before and other personal or academic things. Since these things are different for each student it is hard to use old methods to evaluate them properly. In this project we use a machine learning approach to predict how well students will do in school using data from the UCI Machine Learning Repository. Before we use the models, we clean up the data. Pick the important features to get a better understanding. We try out three methods. Logistic Regression, K-Nearest Neighbours and Support Vector Machine. Compare them. When we tested these methods, we saw that the Support Vector Machine method works a little better when it comes to being accurate while K-Nearest Neighbours gives answers with less work. The main idea behind this system is to help find students who may need help before they start doing. With these predictions teachers and schools can take steps on to get better results. Overall, the results show that machine learning can be a tool, in looking at educational data and can help make better decisions in schools.