الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 73 |  |  |
| | | from | 73 | | |
Abstract
Educational Data Mining (EDM) has the potential to find out possible remedies for students’ attrition, help students’ retention and success at educational environment. It aims at utilizing data mining techniques in educational environments to help students, decision makers, and educators from different perspectives. EDM can use a hybrid data mining approaches combining classification and clustering techniques to achieve high accuracy and predict at-risk students to make extra effort for improving students’ performance.
This thesis aims to propose an efficient predictive model to predict at-risk students’ performance as early as possible to make appropriate intervention to improve the learning process. This is achieved through paying more attention to preprocessing stage and selecting top influencing features that leads to dimensionality reduction which improve the efficiency of the predictive model.
The proposed predictive model achieved an accuracy of 77.26%, F1-Score of 77.8%, and Cohen’s kappa score of 72.0% with the 2-Stages 3-Classifiers and an accuracy of 83.16%, F1-Score of 83.6%, Recall of 83.3%, and Precision of 84.1% after applying DB-Scan.
The performance is evaluated using more than one measure such as the accuracy, Cohen’s Kappa Score, recall, precision, and F1-Score measures for verification and stability. The proposed predictive model outperformed the previous classification models for students’ performance prediction.