In a country where over 23.9 million passengers rely on railways as a mode of travel and daily commutation, it becomes important to keep serving the passengers with inclusive and comprehensive information, to help them make informed decisions. RailYatri is one such initiative that is efforting to simplify train travels, by using deep learning and computer vision. Aiming to be a premier portal for Indian Railways train travellers, it’s a mobile friendly site that answers all train travel questions in a few taps.
At RailYatri, they work to present not just the dry numbers but provide insights behind them so as to help passengers make informed decision about their train travel. Some of the features that it enables are checking PNR status information with confirmation probability, live train status with real time GPS location mechanism, avoiding delayed trains by studying running time statistics, live trip sharing etc.
In a nutshell it is a one stop solution that furnishes an expanse of data-based travel discovery at a passenger’s disposal.
On the idea behind conception of RailYatri-
An idea which was conceived as a data-driven “intelligent” and consumer-centric travel platform, RailYatri helps in reducing the uncertainties in long distance travel by addressing some of the questions like—will my ticket be confirmed? Which is the best train to take? How can I get medical assistance during my travel? And much more.
“The journey has been fulfilling, so to say the least. We have successfully scaled to reach over 15 million app users – being the fastest to reach this scale”, says RailYatri, a startup founded by Manish Rathi, Kapil Raizada and Kapil Saxena.
On being selected for Google’s Accelerator Program-
The team believes that it truly validates the work that team has done so far and the positive impact that is being felt by the users at large. “We have a significant technology overlap in areas where Google has proven products & expertise e.g., big data, deep-analytics, run-time computations based on user locations, AI, machine learning, etc. which is core to how our platform generates its ‘recommendations’”, the company says.
The startup is looking to engage with domain experts at Google on some of the key areas on which the team is currently working on. It is also focusing extensively on scaling consumer products.
On analytics driven product portfolio-
“Our solutions are designed to essentially help travelers throughout the travel lifecycle, and we focus on areas that help save travelers time, money or even anxiety”, says RailYatri”. For instance, their location-driven algorithms can accurately predict the arrival of a train based on the user’s GPS location & train profile and their seat availability predictions have achieved over 97% accuracy. Their app has mappings of equivalent bus & train routes that helps travellers easily switch across modes while doing a seat search.
This startup claims that theirs is the only travel app that can recommend the approach side to the station, automatically show the nearest medical centre en-route your journey, and can facilitate user to check the quality of mobile network connectivity on their route.
On using deep-analytics technology and big data platform that makes intelligent predictions-
The data driven platform by RailYatri makes recommendations to travellers based on actual reported events. The company explains “One can imagine each significant change in status as an “event”. For instance, an event could be when someone cancels a booking, as it allows another waitlist passenger to get a seat. A train’s location & speed can be another event that defines its arrival time at a platform, along the same lines as how Google Maps estimates when you drive a car.
RailYatri uses technology to track the events and analyse the impact to help predict the end outcome of interest to the traveller.
On the big data platform that leverages intelligence and crowd sourced content, RailYatri says “Everyday, over 6 million user GPS data points are analysed to get the best train location and linked to the train profile to generate the ETA at downstream stoppages.”
It further explains that even before the data can be consumed, it has to be first cleaned using multiple levels of filtering to separate the good data from the bad ones. This is where various modelling techniques are used to increase the information reliability using crowd-sourced data from different sources, and it is an ongoing exercise as every event adds to the learning process.
A few use cases-
On being asked to highlight a few use cases where analytics is being used to drive value at RailYatri, the company says that it is using it for instance predicting the platform numbers in advance that can help travellers plan their approach and potentially save on commute, effort and even money if they were to hire a porter. “A lot of travellers have appreciated the seat availability predictions as it helps them avoid spending on the multiple bookings in the hope that at least one would get confirmed”, it says. It further adds that the computations that go into into such calculations are fairly complex.
To take a use case, the waitlist probability is calculated by measuring the impact of a nearly 20+ variables, including traffic patterns of other trains on the same route. RailYatri is solving this and many such complex problems using analytics.
The roadmap ahead-
The company says that is would continue to invest in leveraging crowd-sourced content, analytics based intelligence, user personalization and integrated commerce for a seamless travel experience for their travellers. “We believe that intelligence will define the next generation of travel platforms, and we would like to see RailYatri become the most preferred travel platform for all long distance travellers in India”, it concludes.
Try deep learning using MATLAB