Ixed-ability in EFL crowded classes : problems and solutions

Abstract EN

Student performance prediction has become very popular in Educational Data Mining.

It is used to improve the performance of Students and also improves the quality of the institution In this paper predictive analysis was carried out to determine the class of grades of Students in their final year using their GPA for the first three years of Study.

The data set of Students has been taken from the Faculty of Engineering with 1841 inStanceand 4 attributes.

Five data mining algorithms (Naive Bayes, Random Forest.

SVM, J48, logistic regression) were considered, by using WEKA tools the results show that logistic regression has the best performance to other classifiers .the accuracywas achievedby Fivealgorithms (logistic regrcssion90% ,SViM89% ,Naive Baycs88.1% ,J4887.2% .Random Forest86.9% ) this, data mining creates an opportunity for identifying Students that may graduatewith poor results or may not graduate at all, so that early intervention may be deployed.

.

For future work, this analysis can be further taken forward by using data sets from different size and universities and applying data pre-processing techniques and will be compared and analyzed with other data mining techniques.