CFE 2019: Start Registration
View Submission - CMStatistics
Title: Spatial M-quantile regression with covariate measurement error to model housing price in Milan Authors:  Francesco Schirripa Spagnolo - Università di Pisa (Italy) [presenting]
Riccardo Borgoni - University of Milano-Bicocca (Italy)
Antonella Carcagni - University of Milano-Bicocca (Italy)
Alessandra Michelangeli - University of Milano-Bicocca (Italy)
Nicola Salvati - University of Pisa (Italy)
Abstract: Spatial data have become increasingly common in order to study urban dynamics. However, these kinds of data are often affected by measurement error (ME) due to the bias induced during the data collection processes or for the statistical pre-processing often necessary to estimate the variables of interest at the desired spatial scale. If measurement error is ignored, standard regression estimation techniques may give biased regression coefficients. Moreover, the presence of outliers and influential points in the data can invalidate the assumptions of the classical regression models requiring the adoption of robust methods. A semiparametric M-quantile approach is proposed in order to obtain both bias-corrected and robust estimates of regression parameters. Moreover, this approach allows us to study the differential effect of a covariate at different levels of the conditional distribution of the response variable. The proposed methodology is applied to housing price in Milan, Italy. In particular, the main aim is to study the effect of cultural amenities and related infrastructures on the price levels.