The presence of even a single outlier in a sample estimate can have strong repercussions on the regression models obtained with the method of least squares, nullifying its reliability. This is a condition to avoid in real estate appraisal where regression is used with predictive and explanatory purposes, and therefore it is essential that the regression model best represents the phenomenon investigated. In this study the outliers detection was carried out with a robust regression that uses the method of least median of squared residuals (LMS). With the aid of a special software, the calculations were performed on a sample of houses recently sold in a district of the city of Bari (Italy). The experiment revealed that the regression model, which was initially to be rejected, showed instead excellent performance once all the outliers identified with the LMS were removed from the sample.
LMS for Outliers Detection in the Analysis of a Real Estate Segment of Bari
DE MARE, GIANLUIGI;TAJANI, FRANCESCO
2013-01-01
Abstract
The presence of even a single outlier in a sample estimate can have strong repercussions on the regression models obtained with the method of least squares, nullifying its reliability. This is a condition to avoid in real estate appraisal where regression is used with predictive and explanatory purposes, and therefore it is essential that the regression model best represents the phenomenon investigated. In this study the outliers detection was carried out with a robust regression that uses the method of least median of squared residuals (LMS). With the aid of a special software, the calculations were performed on a sample of houses recently sold in a district of the city of Bari (Italy). The experiment revealed that the regression model, which was initially to be rejected, showed instead excellent performance once all the outliers identified with the LMS were removed from the sample.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
