Real-Time Bus Arrival Prediction: A Deep Learning Approach for Enhanced Urban Mobility

TL;DR

This research introduces an innovative, AI-based, data-driven methodology for predicting bus arrival times at various transit points (stations), offering a collective prediction for all bus lines within large metropolitan areas through the deployment of a fully connected neural network.

Abstract

In urban settings, bus transit stands as a significant mode of public transportation, yet faces hurdles in delivering accurate and reliable arrival times. This discrepancy often culminates in delays and a decline in ridership, particularly in areas with a heavy reliance on bus transit. A prevalent challenge is the mismatch between actual bus arrival times and their scheduled counterparts, leading to disruptions in fixed schedules. Our study, utilizing New York City bus data, reveals an average delay of approximately eight minutes between scheduled and actual bus arrival times. This research introduces an innovative, AI-based, data-driven methodology for predicting bus arrival times at various transit points (stations), offering a collective prediction for all bus lines within large metropolitan areas. Through the deployment of a fully connected neural network, our method elevates the accuracy and efficiency of public bus transit systems. Our comprehensive evaluation encompasses over 200 bus lines and 2 million data points, showcasing an error margin of under 40 seconds for arrival time estimates. Additionally, the inference time for each data point in the validation set is recorded at below 0.006 ms, demonstrating the potential of our Neural-Net-based approach in substantially enhancing the punctuality and reliability of bus transit systems.

Figures (8)

• Figure 1: Quarterly Public Bus Transportation Ridership in the U.S. In 2020 and 2021, public transportation ridership was less than half its pre-pandemic level. While bus ridership has recovered somewhat, it was much lower in the second quarter of 2022 than in the final pre-pandemic quarter. Bus ridership for commuters grew by 66% in the second quarter of 2022 mallett2022public.
• Figure 2: Integrating an AI prediction model into a mobile bus app enhances user experience and operational efficiency. Our model predicts bus arrival times using diverse data sources, providing real-time precision. Users can easily access these predictions via cloud-based services for a reliable travel experience.
• Figure 3: Average delay among all bus lines. Initial analysis of records in the New York dataset shows that the average delay across all bus lines equals 491 seconds.
• Figure 4: Structure of our model based on the Fully Connected Neural Network. One-hot encoding applies to bus lines and extends it to 232 features. These converted features with other 5 features, including distance, day type, rush hour status, bus stops, and far status feed to the Fully Connected Neural Network. The proposed model consists of 5 hidden layers and ReLU as an activation function. The number of neurons in each hidden layer can also be seen in the figure. H1-320N indicates that the first hidden layer consists of 320 neurons.
• Figure 5: Performance of the model for all bus lines. The average RMSE across all bus lines in the validation set is 35.74 seconds. Among the prediction error values, bus line 160 has the greatest RMSE, with a value of 119.99 seconds. In contrast, the lowest error belongs to bus line 76 with an RMSE equal to 12.42 seconds.