Abstract:
The bus location data received by Chartr from DTC does not have information about the route that buses are on at a particular time of the day. This information creates a major hindrance in unifying the data with the location data from DIMTS. Further it disables us from solving many further problems such as trip planning and Seamless Ticketing. The current route detection algorithm for DTC buses has a high percentage error of around 38 percent. The objective of this project is to improve the accuracy of route detection by devising different methods to detect routes and comparing the results obtained by the various approaches as well as with the true value of the bus route which is obtained from the user ticketing data. We use static techniques such as comparing with bus schedule charts, as well as LSTM modelling of routes for route classification. In the second part of the research project, we create a portal for visualising Delhi pollution data, collected using ATMOS sensors placed on Delhi buses. Furthermore, this pollution data is used to create a PM2.5 pollution prediction system, using an LSTM based forecasting model. Keywords