A Machine Learning Method for the Analysis of Urban Italian Mobility

Abstract

The aim of this paper is to consider a method of machine learning to analyze the problem of the sustainability of the urban transport in Italian cities. First of all we recall the definition of sustainable mobility then we present some indicators considered in our analysis. The methodology used in this paper are decisional trees. We both consider classification and regression trees. We have chosen two different dependent variables one for classification trees (a categorical variable: Macroregion according to NUTS 1: North West, North East, Centre, Islands and South of Italy) and one for regression trees (a quantitative variable: PM10 maximum number of days in excess of the human health protection limit foreseen for PM10). In order to test the performance of this methodology we have applied random forest. The analysis has been performed using SAS language.