This presented research paper mainly studies the frequent itemsets mining approach for finding the most important attribute to overcome the existing problems in the extraction of relevant information by using data mining approaches from a huge amount of dataset. Firstly a state of art diagram for prediction is designed and data mining classifier like naive bayes, support vector machine, decision tree, k- nearest neighbour are compared and then proposed methodology with new techniques are proposed. Moreover, a new attribute filtering association frequent itemsets mining algorithm is presented. Then, by analyzing the feasibility of the proposed algorithm, the data mining classification classifier is compared. As a result, SVM produces the best result among all the classifier with attribute filtrating and without attribute filtrating. With attribute filtrating algorithm enhances the accuracy of all the other classifier.