Prediction Accuracy of Academic Performance of Students using Different Datasets with High Influencing Factors

Abstract: For the past few years, a lot of higher educational institutions tussle for providing quality education. To provide a quality education the institutions need information regarding their students. In higher education institutions a substantial amount of knowledge is hidden and need to be extracted. Various Data mining techniques are useful for deriving such hidden knowledge. The knowledge can be any student specific information like academic performance, dropouts, course preference, subject specialization etc. The quality of the students in a higher education institution is categorized by their academic performance. Many socio economic, non academic and academic factors influence the performance of the students. The factors that describe student performance can be used for predicting students performance by using a number of well - known data mining classi?cation algorithms such as ID3, Simple CART, J48, NB Tree, MLP, Bayesnet etc. The model is mainly focused on finding the prediction accuracy of academic performance of students using two different datasets. The experimental model also proves that the student attributes considered are highly influential in predicting the results using MLP classification algorithm.

Keywords: Educational Data Mining, Academic Performance, Prediction, Classification, Influencing Factors.