Student Performance Prediction Using Machine Learning

Predicts at-risk students from attendance, assignment, and test-score data.

Many students fall behind quietly. By the time low marks show up in an examination, it is often too late for a teacher to help, although the warning signs such as falling attendance, missed assignments and weak class tests appeared weeks earlier. This project uses machine learning to predict which students are at risk so that teachers can act early.

The project uses attendance, assignment and test-score data, cleans it, handles missing values and engineers useful features such as attendance trend and score improvement. It then compares regression models that predict a final score with classification models that predict pass or at-risk, including linear and logistic regression, decision trees and random forests, using cross-validation and measures such as accuracy, recall and the F1 score. Recall for the at-risk group is emphasised because missing a struggling student is worse than a false alarm. A Flask dashboard lets a teacher upload class data, view predicted risk for each student, sort by risk and see which factors contributed most.

You will learn data cleaning, feature engineering, model comparison, evaluation on imbalanced data and how to present predictions responsibly, including the ethics of labelling students. The Project Reference Guide explains every step and the Reference Implementation includes the notebooks, trained model and dashboard.