Comparative Analysis of Encoding Methods in Regression Predicting Models of Bachelor's Final Marks

This study assesses the impact of One-Hot and Target Encoding techniques on the accuracy of predicting bachelor's degree final marks in economics and management. Employing regression models across six machine learning algorithms -Linear Regression, Support Vector Machine, Decision Tree, Random Forest, XGBoost, and Neural Network- we analyze a comprehensive dataset from Hassan II University of Casablanca. This dataset includes both demographic and academic performance data of students. We focus on the application of encoding methods to categorical variables and evaluate model performances based on Mean Squared Error (MSE) and Mean Absolute Error (MAE). Our findings highlight the differential effectiveness of encoding techniques in enhancing the precision of predictive models for academic outcomes.