Deep Learning-Based Correlation Analysis of Public Transit and Ride-Sharing: A Study of Urban Mobility Patterns in Paris

Abstract

Urban transportation systems have evolved significantly, particularly with the rise of private ride-sharing services. In Paris -as our case study-, urban mobility is shaped by the interaction between public transportation systems and services like Uber. This study analyzes and predicts the relationship between these two transport modes using a Convolutional Neural Network (CNN) model. By processing the spatio-temporal heatmap visualizations from Uber ride-sharing activity and public transport usage, the CNN model predicts the correlation coefficients, dynamically capturing how these two transportation modes interact. This analysis reveals how ride-sharing services either complement or compete with public transit at different times and locations in Paris, offering urban planners valuable insights into optimizing transportation systems. The CNN model achieved a mean accuracy of 93.31%, predicting varying correlation coefficients during off-peak and peak hours. The results suggest that ride-sharing supplements public transport in underserved areas during off-peak hours but competes with it in central areas during peak times. This predictive correlation analysis provides valuable insights for urban planners, helping optimize resource allocation and improve the integration of public and private transport, contributing to more efficient and sustainable urban mobility solutions across diverse geo-locations.