Abstract - Machine learning (ML) is transforming education and fundamentally changing teaching, learning and research. The ML technique helps the institution to utilize the resources in better ways and produces results in the best possible effective manner. The learning combines various processes like data preparation, classification, association, building models, training, clustering, prediction etc. to improve performance of students. It helps to the students to select any particular course based on their choices and previous performances. The main focus of this study is to analyze the various classification techniques over the educational data. The comparative study was conducted to predict the student performances based on some social variables (extracurricular activities, family education support, and desire for the higher education), previous exam grades and along with other attributes. In this paper, the Naive Bayes(NB) , Bayes Network(BN), Radial Bias Function (RBF), Multi-Layer Perceptron (MLP), Back Propagation Network(BPN), Random Forest(RF), J48, Radial Basis Function Network (RBFN) classification techniques were chosen for the experiment. After testing all the data over the mentioned classification we found that correctly classified instances percentage is 100% for Random Forest and it is highest compared to other classification algorithms.