The Use of Quantile Regression in Data Analysis Tasks

Abstract

Theoretical and applied aspects of the use of quantile regression in data mining tasks are considered. The features of application of quantiles for modeling and data mining are introduced. It is shown that the most complete picture of the impact of the explanatory variable on the shape of the distribution provides the finding of conditional quantiles, that is, the use of quantile regression. The methods of construction of quantile regression are analyzed and the two-step procedure with the use of the method of k-nearest neighbors for the computation of the quantile functions and kernel smoothing for the final quantile regressions is carried out. The analysis of the efficiency of application of quantile regression models in different socio-economic areas is analyzed.