How Google Maps predicts traffic accurately in India

• Google Maps has enabled the ability to track traffic delays in several cities across the world.
  
• Traffic delays data is updated in real-time thanks to a wide range of data input sources.
  
• The Machine Learning model employed by Google uses several variables to deliver these results.
  

Google Maps has become a quintessential tool in our daily lives. Whether it is finding the best route possible to reach your destination, or to locate a good hotel in your vicinity, Google Maps has all the tricks up its sleeve. Google Maps’ route predictions and commute timings are accurate as well, allowing people to rely on it.

However, have you ever wondered how Google Maps manages to do it? Predicting travel times, especially amidst city traffic is not an easy task. It’s not just traffic either – often, unforeseen delays caused due to accidents or a road being dug up can further delay commutes.

Despite years of experience, we can often go wrong with our predictions, and yet, Google Maps’ traffic predictions are accurate more often than not.

So, how does Google do it?

The answer – Machine Learning.

Google’s new traffic and bus delay prediction system went live a few months ago in India. To make sure its predictions work well, Google says it extracted training data from sequences of bus positions over time, as received from transit agencies’ real-time feeds. This data was then integrated with car traffic speeds on the bus’s path.

The data obtained includes details like the bus route, trip location and timing. Turns, streets and stops are also a part of the data obtained by Google.

Each part of the trip is split into timeline units and Google makes forecasts based on these units. Once this is done, Google reports the forecast based on the query of the user – it can be just one unit or a sum of all the units.

Tuning the model to the local beat

Each city moves differently, and Google recognizes this.

Ever miss a bus due to a 5-minute delay? We have all been there at one point or the other. To make sure its bus predictions are accurate enough, Google uses data obtained from traffic signals and rush hour crowds at bus stops. Additionally, the commute times can vary wildly on the same route at different times of the day.

To wade through these issues, Google says that it puts time in a 4-dimensional loop, which is demonstrated via the rendering above.

In total, Google combines all these widely different signals and combines them together to address user queries based on several variables.

The traffic delays feature is live in Google Maps in several cities in India. We tested it in Hyderabad, Bengaluru and Delhi, and the predictions are more or less accurate. When there is a change in the route, Google updates the estimation in real-time.

SEE ALSO
9 incredibly useful Google Maps features everyone should know about
Google Maps is working on tool to tell you how crowded the buses and trains are
How to change your route on Google Maps on desktop or mobile, to choose an alternate route or make your own