Household Vehicle Ownership Prediction Using Machine Learning Approach

Abstract

Predictions of vehicle ownership and their influencing factors play an important role in transportation policy making. In the era of urbanization and globalization, vehicle ownership patterns have become a more relevant issue in developing countries attempting to achieve sustainable transportation development goals. Machine learning techniques facilitate tremendously in predicting the vehicle ownership patterns, required to achieve the above said goal. In the proposed work, the machine learning models such as decision tree, random forest and multinomial logistic regression models are applied over household datasets, to predict the household factors influencing vehicle ownership. According to the ML model, influencing factors on Household Vehicle Ownership (HVO) prediction are number of persons having DL in a household, number of persons drive in a household, number of persons using ride source service or public transport in a household. Analysis of datasets showed that, total income of a household, number of persons in household and distance travelled each day are positively associated with vehicle ownership of the household. The predictions of this study would be useful mainly for local governments, transportation agencies, planners and policy-makers.