Evaluating student performance is very difficult task for academic area. The educational performance play a vital role in classifies the student in higher education. The student performance effect various factor like learning process, personal and social. This paper demonstrates the impact of student positive or negative performance of student success. Here the most commonly used prediction algorithms were LWL, random forest and bagging. After apply this three algorithm we present a novel model combination of this three algorithm name as ensemble. With the help of data mining algorithm we predict the student dropout rate, which is helpful for academic progress. For this we have to collect big and authentic data which can be done through uniform district information system for education DISE).