A weighted quantile approach to improve prediction of graduate student university career

This paper is a pioneering attempt to apply the weighted quantile regression method to college admissions in order to consider the controversial behavior of the final mark of high school diploma (HSD). The new approach takes into account the fact that student characteristics have a varying impact at different part of the indicator of college performance (here: weighted average mark or WAM), that is, for particular quantiles. The main advantage of this method is the ability to handle models in which data exhibit dissimilar levels of quality. Furthermore, quantile regression is also a robust technique to outliers in the response variable. Our results reveal that there exists a complementary correlation between HSD and WAM and between WAM and the average grade of the first term of the last year at high school (AGFT)