Educational data mining aims at identifying hidden patterns in the field of education in order to improve the learning process from the data collected from educational institutions. This paper proposes a model which employs SVM and Regression analysis on data collected from final year undergraduate students of various colleges affiliated from GGSIP University. Data has been collected with the questionnaire based on 27 non intellectual correlates identified from 6 unique spheres namely demographics, personality, motivation, self regulatory learning strategies, student�s approach towards learning, and psychosocial contextual influences. The reliability of the questionnaire has been checked by calculating the cronbach alpha value which is significant. To test the significance Chi square test for goodness of fit is applied. The paper also provides the meta analysis using random � effects model. It is shown that SVM provides better results in predicting performance of students than Regression analysis.