Arrival Time Accuracy Research
Providing accurate, real-time bus location information reduces riders’ wait times and thus the length of their daily commutes, leaving more time for vocational, educational, and leisure activities. Our SmartSAT mobile web app will track bus location using a combination of Google APIs, Google Live Transit Updates service, and bus Global Positioning System (GPS) to improve the prediction accuracy of bus arrival time.
Proposed Work
Existing approaches use GPS to collect real time location information of a vehicle to predict public transport arrival times on bus stops. However, the evaluation of the proposed algorithm shows errors in predicted times. In addition, issues related to GPS outages and location errors have not been addressed. To address such issues, other approaches used the historical and real time location data to determine the current location of a bus and to predict arrival time at a bus-stop. Results, however, show mean error of 5 minutes in the arrival time prediction. To improve the arrival time accuracy, we will use information provided by Direction API and Distance Matrix API of Google Maps, Google's Live Transit Updates service, and Bus GPS to calculate bus travel times based on real-time traffic data to obtain accurate real-time traffic data, compute travel times and predict the arrival time at a bus-stop. Data analytics will then be used to assess the accuracy of the predicted arrival time by comparing it with the actual bus arrival time. The study will also assess the accuracy of the current scheduled time by comparing it with the actual bus arrival time.
Methodology
The primary outcome measure that will be used to evaluate our hypothesis is the arrival time prediction error (i.e., the difference between the actual and predicted bus arrival times). The outcome will be evaluated using multiple-trip data collected over the six VIA bus routes selected for this project. We will measure the mean error (difference in time) between the actual arrival time and the predicted one. Results will show the accuracy of predicted times, and whether this accuracy can be improved. If the accuracy is improved, we will determine the accuracy of the scheduled times and if they need to be changed.