Predictive Analytics for Car Dependence: A Machine Learning Approach to Influence Travel Behavior

Abstract

Travel behavior is vital for producing effective interventions because car dependence is one of the main barriers to achieving sustainable urban mobility, which highlights the importance of a deeper understanding of the contributing factors. Understanding and altering driving behavior: analysis with ML Predictive analytics enable to forecast behavioral trends and help to change driving behaviors. We applied various machine learning methods including random forests, gradient boosting, and neural networks, to predict individual travel Behaviors and to find key determinants of car dependence. These designs resulted from using a comprehensive data set that revealed travel survey results, GPS data, and demographic information. This dataset included factors such as travel time, costs, availability of public transport, and city density. To determine the effectiveness of the models, we measured accuracy, precision, recall, F1-score, and area under the ROC curve (AUC-ROC) for each of the models. The models accurately predicted car dependence with 85% accuracy (AUC-ROC 0.89), confirming that they were able to maintain predictive power. Ensuring access information on public transport and urban density were the key factors that were indicative in our analysis of feature importance. Moreover, we designed a personalized travel planning intervention informed by our predictive models, which ultimately reduced car usage by 15% and increased ridership of public transport by 18%. These findings highlight the capacity of machine learning and predictive analytics to provide deeper insights into car dependence, and to guide targeted interventions that promote sustainable travel in urban environments. Future studies will be directed towards investigating the long-term effects of such interventions and integrate other factors for example, social norms that may impact travel behavior.