Schools are considered as the backbone for long-term economic progress. No country can develop without increasing their education level. Despite the fact that the Portuguese population shows a brilliant development in their educational level from last decade, but still Portugal lies on the tail surrender of Europe in statistics because of excessive levels of student failure. Primarily, this costs a lot better in the middle of the elegance of Mathematics and Portuguese. On the other hand, the field of data mining (DM), the purpose of extracting the high-stage knowledge of raw statistics, automatic gear compelling offer to a useful source of training domain. This paper pursues to improve the overall performance of middle school students of Portugal through two variables decision tree, which is a favorable approach to data mining used for classification, prediction and factors explored with the help of their significance. Results shows that, provided the first and / or second interval school grades, awesome prediction accuracy can be achieved. Despite the success of students strongly influenced by father's job assistance; evaluation has clearly shown that there are also other elements (such as learning time, mother's occupation, the desire of higher education, the paid-classes and the travel time from home and school, etc.) are important elements which have great impact on the performance of students in secondary school education in Portugal. As a direct result of this study, through which specialize in these factors and create a kind of policy is mainly based on studies in the country width exceptional level of education may increase at the secondary level that produces goose bumps to the stage of higher education in Europe.