Assessment of Satisfaction Level for Existing Public Transport Systems using Machine Learning: A Case of Bhopal (India)

Abstract

Most people are now highly dependent on private motorize travel because of the low Satisfaction level of public transit modes. This leads to poor public transit share in most of the cities in India. This study is intended to identifying the influencing parameters; those affect the satisfaction level of a PT mode, and also develop a model for assessing the Satisfaction Level of these public transport modes in Bhopal, India. A Machine Learning-based approach is used to analyze 1189 responses from the user. It classifies data in various dimensions that shape customers satisfaction among each mode of public transport (i.e. BRTS, Mini-Buses and Magic-Van service) and develops a model for evaluating the overall satisfaction of these PT modes in Bhopal. An On-board opinion surveys were done for identifying the parameters which influence the Satisfaction of these PT modes and also develop a model for evaluating the Satisfaction of these modes. Based on the literature, Delphi survey and opinion survey, eight parameters have been identified that influence the satisfaction level of these modes in Bhopal. Further by correlation matrix, most influencing parameters (key parameters) were considered amongst them for evaluating the satisfaction level for these modes. To determine the coefficient values acting from each of the respective elementary scores, we used a trained linear regression, multilinear regression, considering the classification of customers who assessed the performance of their satisfaction. This model will help to increase the Satisfaction of these PT Modes which result to increase the transit ridership in Bhopal. Adopted methodology in this study can help decision-makers to improve public transport services so that transit ridership can be improved.