Route Recommendor system for KSRTC

This proposal broadly aims at the adaptation of technology to improve the operation of KSRTC bus routes so that theearning per km is maximized and the reliability of operations is ensured. A pilot project was implemented for Kerala State Road Transport Corporation (KSRTC) to analyze the routes in terms of the revenue and the passenger traffic and

• Development of algorithms for bus and crew scheduling based on the above analysis.
• Development of software for the simulation of bus traffic in all proposed routes.
• Development of software for bus scheduling and crew scheduling.
• Initiation of new circular routes in the city and the assessment of the performance in terms of earning per kilometer (EPK).
• Implementation of a Hop on traffic system within the city by coupling the various circular routes and the assessment of performance.
• Initiation of new routes, based on passenger survey and the assessment of the performance in terms of earning per kilometer (EPK).

The passenger traffic of selected routes over nine months is averaged and plotted to study the traffic model.

• The Markovian behavior of the passenger data is revealed by considering the passenger statistics in a route. The passenger data Xn depends only on Xn1 and not on the set X0, X1, ..., Xn2.
• The different states in the Markov model indicate the total number of seats occupied by the passengers in the bus.
• A deep learning-based LSTM architecture is designed to extract the features of bus passenger data.
• The predicted data from the LSTM architecture is given as input to the particle filter to correct the errors.
• The experimental setup to implement this prediction is as shown in item