Authors: Narges Rashvand, Sanaz Sadat Hosseini, Mona Azarbayjani, Hamed Tabkhi

Published on: March 27, 2023

Impact Score: 8.35

Arxiv code: Arxiv:2303.15495

Summary

• What is new: An AI-based, data-driven methodology for predicting bus arrival times using a fully connected neural network, significantly reducing the error margin in arrival time estimates.
• Why this is important: The mismatch between scheduled and actual bus arrival times in urban settings, leading to delays and decreased ridership.
• What the research proposes: A neural network model that predicts bus arrival times at various stations for all bus lines within large metropolitan areas, using historical bus data.
• Results: Achieved an error margin of under 40 seconds for arrival time estimates with an inference time below 0.006 ms for each data point, across over 200 bus lines and 2 million data points.

Technical Details

Technological frameworks used: Fully connected neural network

Models used: Not specified

Data used: New York City bus data

Potential Impact

Public transportation authorities, urban planning departments, and transit app developers could benefit from this research.

Want to implement this idea in a business?

We have generated a startup concept here: TransitAI.