Dataset for Advanced Public Transportation System under Limited Infrastructure Support
Description
We present the dataset of bus trajectories and commuters' activity records (collected inside the bus) for the trips conducted in the Ahmedabad-Gandhinagar district of Gujarat, India. It also contains the trip demand for the Ahmedabad-Gandhinagar district of Gujarat, India. The data can be used for arrival time prediction, automatic bus-stop detection, and bus crowdedness detection for the scenarios of developing countries where Intelligent Transportation System (ITS) infrastructure deployment is limited. The automatic bus-stop detector resolves the problem of dynamic bus-stops on the routes. Moreover, the arrival time predictor can disseminate the estimated bus arrival time to the waiting commuters, and the bus crowdedness information can be utilized for scheduling and provisioning buses in crowded areas. The data is collected using the android application, which periodically fetches the GPS records at the frequency of 1 Hz and the accelerometer records at the frequency of 40 Hz. It processes the bus trajectories to calculate arrival time prediction and all en-route stoppages of the bus trips for automatic bus-stop detection. However, it is challenging to infer whether a bus stoppage is due to bus-stops, traffic jams, or junctions. We observed that the bus-stop is a stoppage where commuter boards the bus. Therefore, the bus trajectory data and the commuters' records are opportunistically processed using the baseline classifier comprising of stepping detector, transport mode classifier, and commuter activity classifier to determine the bus boarding events and commuter's state (as standing/sitting) after boarding the bus. The software filters the en-route bus-stops using the bus boarding locations. The commuter's state (standing/sitting) is used to determine the crowdedness inside the bus. The sitting commuter's state indicates the availability of seat and a lesser level of crowdedness. In contrast, the standing commuter's state shows that the commuter did not got a seat on boarding the bus and reflects a higher crowdedness level. The software implementation uses open-source libraries and is reproducible. The link for the reproducible capsule of the implementation can be found in the related link section. The dataset information and meta-data for different records is mentioned in the respective folders.
Files
Steps to reproduce
The application is created for the smartphone operated using the Android operating system to collect the data. The application fetches the GPS records at 1 Hz and the accelerometer records at 40 Hz.