Deep Learning Models for Predicting Public Transport Delays

The purpose of this study is to quantify the efficiency of Deep Neural Networks (DNN) in predicting delay in public transport and enhancing the efficacy of scheduling. The precision of the DNN model is contrasted with that of the conventional Support Vector Machine (SVM) model to comprehend which method makes more accurate predictions of delays and enhanced scheduling. Materials and Methods: There are two groups in this study. Group 1 is a Deep Neural Network (DNN) model, comparing different data inputs like weather, traffic, day of the day, and historical transport delays using a sample size of 1000 delay points. Group 2 is the mention of a Support Vector Machine (SVM) model comparing the same data using a different prediction algorithm using the same sample size of 1000 delay points. G Power value = 80% with threshold = 0.05% and confidence interval = 95%. Results: The Support Vector Machine (SVM) model is less accurate than the Deep Neural Network (DNN) model in public transport delay prediction. Delay prediction accuracy for the DNN model ranges between 85.5% and 92.3%, whereas that of the SVM model ranges between 74.0% and 82.1 %. The optimal performance of the DNN model is captured within an hour of the peak rush hours and at a significance level of approximately 0.0042. The enhanced predictions resulted in enhanced scheduling, which enhanced the overall transport efficiency. It was concluded in this work that the Deep Neural Network (DNN) enhances the accuracy of public transport delay predictions much better than the Support Vector Machine (SVM), which ultimately results in enhanced scheduling and the performance of the public transport system.