Feature Selection for Performance Prediction using Decision Tree

Abstract

With the proliferation of digitalization from past decades, there has been exponential growth in data. The data is considered as the new oil. Useful insights can be extracted from the data and can be used for the growth of industry or organization, the branch of computer science that deals with the discovery of novel information and insightful pattern from the raw data is called Data Mining, it is used almost in every field from banking, healthcare to entertainment and surveillance. Here, specifically, the paper discusses about the data mining used in the field of education called Educational Data Mining (EDM), it is an inchoate data mining research area that aims to improve the students' performance, provide quality education, helps students to determine career choices, etc. This research paper aims to describe feature selection criterion and how combination of features like internal marks, mid-semester marks, etc. helps in determining students' performance using decision tree classification method. The results obtain can be used by instructors and teachers to plan structured approaches for the students performing low in their academics, need attention and counseling from their tutor guardians. Thus, with early prediction action can be taken within time to improve the result of students and the overall performance of an institution.