Data Mining in Educational Systems for Effective Student Mentoring

In an educational institute mentoring of students is an essential feature to provide equitable services to students having a varied background and to minimise dropout rates. Conventionally, the mentors seek for student data manually, from files and records with no assistance from an automated system. To cope up the worsening student teacher ratio and the exponentially increasing data, assistance from an automated system becomes significant. Through this paper, we suggest the integration of a Machine Learning driven computer program, with mentoring in order to ensure an effective guidance system. In the proposed system, Educational Data Mining (EDM) was used to create a regression based prediction model to aid professors in the mentoring process by generating an estimate of students' future performance. To make this system more accurate, aberrant factors affecting student performance like travel time, type of residence, gender, caste, parents education and working status, along with conventional academic and analytical features like CGPA and extra-curricular were used to train the Machine learning model. After implementation, it was observed that the regression model built using Support Vector Machines (with RBF Kernel) had the least square error. This model was then integrated with a mentoring system to build a computer application to assist professors. This application was rife with features like attendance and session tracking, graph generation and CGPA prediction. With the assistance of the application, mentors can better create a tailor-made plan for individual student and further enhance the effectiveness of student mentoring.