Predicting Dropout Student from School

Abstract

Predicting student dropout at school has become a difficult challenge due to both the high number of factors that can affect the low performance of students and the imbalanced nature of these types of data sets. In this paper, different data mining approaches are proposed for solving these problems using real data about 858,013 school students from 34 districts, of Baluchistan province. We use in our experiments school census data of the year 2015-2016. Firstly, we select the best attributes in order to re- solve the problem of high dimension. Then, rebalancing of data and information gain and correlation based for feature selection techniques are applied as well as different classifications models has been applied in order to resolve the problem of classifying imbalanced data. We model this problem with class values dropout and non-dropout. We also propose to use a classification prediction and accuracy of each approach are shown and compared in order to select the out- comes of each approach are shown and compared in order to select the best to improve classification accuracy, specifically with regard to which students might dropout.