Bus Alighting Stop Estimation Using Trip-chain and Bayesian Methodology

In order to accurately predict the situation of passengers getting off the bus at a bus stop in the future, prediction models based on different Bayesian methods are proposed. Each model can be divided into two stages. The first stage is to predict which bus stops need to get off for a partial ride based on travel chain assumptions. In the second stage, three models based on the naive Bayes model, the semi-naive Bayes model and the Bayesian network are sequentially trained according to the prediction results of the first stage. The three models were tested and validated using data obtained from multiple bus routes at specific locations. The test results show that, compared with the prediction cases based on the travel chain assumption, the improved models have better prediction accuracy and are more suitable for the randomness of the real situation. In addition, in the comparison of the three models, the prediction model based on Bayesian network has better predictive ability in general, and can more accurately predict the time of passengers getting off the bus in real situations, and its prediction results are more realistic, followed by Predictive models for semi-naive Bayesian and Naive Bayesian models.