Clustering of secondary school students in Portugal

The dataset about the secondary schools in Portugal has been handled in the paper. Nowadays data analysis and mathematical statistics methods allow researchers and staff of universities to understand hidden dependencies in the data about students. In the original data competition for which the handled dataset was presented the main goal was to explain the final exams grades by means of social and behavioral parameters of a person. In the paper this question is researched in a new way. The clustering technique allows dividing students into a few groups. Mathematical models of the final grade are special for each cluster. Thus, models achieve some kind of individuality saving generality. Comparison of results of models constructed for the whole dataset and for each cluster has been prepared. Such data analysis technique can be implemented to handle another datasets with different set of features. Obtaining results of data analysis the staff is able to make conclusions on individual way of dealing with every cluster or students and some clusters can be analyzed in individual manner.