Comparing the performance of FCBF, Chi-Square and relief-F filter feature selection algorithms in educational data mining

Educational Data Mining (EDM) is a very vital area of Data Mining, and it is helpful in predicting the performance of students. This paper is a step towards identifying the factors affecting the academic performance of the students. It is very necessary to increase the quality of dataset as to get better prediction results. There are many feature selection algorithms, however three filter feature selection algorithms FCBF, Chi-Square, and ReliefF are selected due their better performance, and applied on three different student�s data sets. The results of three filter feature selection algorithms are evaluated. The result of the paper extracted that Chi-Square and ReliefF perform better than FCBF on a dataset with larger number of features, however the performance of three selected algorithms is found worst on a student dataset with less number of instances. The analysis of the results of algorithms shows that, Home to Institution (School/College) travel time is one of the important feature affecting the performance of the students. Student�s previous academic background and Socio-economic factors also appeared to be the important factors for predicting the academic performance of the students. © Springer Nature Switzerland AG 2019.