Preliminary Investigation of Women Car Sharing Perceptions Through a Machine Learning Approach

Abstract

Mobility studies have shown that travel patterns and means use vary a lot comparing women and men behavior. In recent years, new solutions have been introduced in the urban mobility offer and the interest raised in investigating how they can help in reducing the gender mobility gap. The current work analyzes 2934 responses collected through a car sharing survey proposed in Italy with the precise objective of considering women and men like different kinds of users to delineate characteristics that could influence car sharing modal choice. A hierarchical clustering technique is applied to the dataset collecting a selection of questions, mainly focusing on socioeconomics features, travel patterns and individual habits. The algorithm identifies 8 clusters in the male dataset and 9 clusters in the female one, defined according to characteristics aggregating the survey respondents. Thus, a selection of these groups of respondents is analyzed in more detail according to their percentage of car sharing users, also comparing the results among male and female datasets. Many common attributes are found in clusters irrespective of the gender, showing how the interest (and its lack) toward this service affects women and men similarly. At the same time, this analysis helps in identifying the features characterizing the users to investigate how this new mobility offer can help in reducing the gender mobility gap.