dc.description.abstract |
Many data mining techniques proposed to extract knowledge from large data in different fields. Data mining is important in educational institutions to extract knowledge from educational data. This study aimed to predict students’ academic performance and explores factors assumed to affect students’ performance in English language and Mathematics subjects. Hence, we propose a student performance prediction model using data mining techniques with the desired attributes which are collected based on student’s learning behaviors and students background. R tool is used for undertaking the experimentation. For the conducted research for its relevance depends on our data property we used Knowledge Discovery Databases (KDD) Process. The researcher adopted three data mining algorithms including: Decision Tree (J48 classifier), Naïve Bayes algorithm, and Random Forest were used for classification purpose and we used percentage split testing option to check the accuracy of the classifier and compare the performance in the model building process based on precision, recall, sensitivity and specificity rates. The experiment result shows that in the classification of student in a given performance index the accuracy of J48 classifier is a little bit higher than Naïve Bayes classifier and Random. The result of the classifier accuracies are 72.2%, 72.4 %, 69.23 % respectively. |
en_US |