Students Performance Prediction by Using Data Mining Algorithm Techniques

Abstract

The use of educational data mining with different techniques aims to solve various problems of the educational environment, especially educators to help learners avoid failure in the study. In this proposal, information on extracting educational data and how to extract knowledge from it to help the educational environment is presented. In this context, previous studies aimed at predicting student performance and the traits that influence their performance have been extensively presented in this research. However, each researcher used different attributes with different techniques to reach the same goal which is to predict the level of students based on data collected from educational institutions. In the context of data preprocessing, the selected data is worked on and configured to work correctly with WEKA. Then, in order to better understand the data, data attributes and data format were introduced. In addition to that we propose in our study to use a learning dataset from the UCI repository and analyze it using Waikato Environment for Knowledge Analysis (WEKA), classification and regression techniques were applied. Then, the characteristics that affect the student's performance, and predict the student's academic performance on the other hand, were determined. Finally, the Random Forest algorithm showed a superiority over the Linear Regression algorithm in the prediction of values only in training set mode.