<![CDATA[Predicting final exam scores is crucial for evaluating and improving academic performance. This study applies the linear regression method in data mining to predict final exam results based on factors influencing student performance. The methodology employed is the Cross-Industry Standard Process for Data Mining (CRISP-DM), comprising business understanding, data understanding, data preparation, modeling, evaluation, and deployment. This study utilizes the public dataset student_performance.csv, consisting of 395 records and 33 attributes. Through an attribute selection process using a correlation matrix, 7 predictor attributes with the most significant impact on the label were selected. The data mining process was executed using RapidMiner. This research yielded the linear regression equation model : y = 0,729 – (0,024 x medu) – (0,020 x Fedu) + (0,053 x failures) – (0,077 x gout) – (0,012 x absences) + (0,126 x G1) + (0,862 x G2). The model performance evaluation resulted in a Root Mean Squared Error (RMSE) of 0.675. Based on these results, the developed linear regression model demonstrates good accuracy and is highly recommended for predicting student exam scores.]]>
Copyrights © 2026
