The real fact in the education institute is the significant growth of the educational data. Data mining techniques are used to extract the useful information and to predict the student academic performance. The main aim of this paper is to construct predictive model for student academic performance. As there are many classification techniques are available, in this paper naive bayes classification technique is used. This paper presents and analyses the experience of applying certain data mining methods and techniques on student data in order to prevent academic risk and desertion.