Precognition of Students Academic Failure Using Data Mining Techniques

Abstract: This paper proposes to pre-recognize student�s academic failure. Real time data on school or graduating students from an institute is taken and various data mining techniques (classification algorithms), such as induction rules, decision trees and naive bayes are applied on it. The results of these algorithms are being compared and optimized for foretelling which students might fail in future. We first consider all the available attributes of students, then select few best attributes and finally, rebalance the data using classification algorithms. The use of data mining concept in the field of education is called as Educational Data Mining, EDM [2]. This paper focuses on designing various methods that will help the teachers and the principal (Administrator) of the school to figure out the weak students and improve their educational standards and environment in which they learn. I propose the use of data mining procedures, because the complexity of the problem is high, data to be handled is very large and often highly unbalanced. The final objective of this paper is to detect the failure of students as early as possible to prevent them from dropping out and improve their academic performance. The outcomes are compared and the best results are shown.

Keywords: Data Mining, Educational Data Mining, Decision Trees, Induction Rules, Rebalancing Data, Classification Algorithms