ON-DEMAND
RIDESHARING
PLATFORM

Matching Demand and Supply
using Machine Learning

Know More

About Cab2Share

Cab2Share is a high-tech ride-sharing product with a goal to offer on-demand, on-time, economical and friendly rides to the riders while maximizing the earnings of driving partners. Our cutting-edge algorithms, massively scalable data platform and newer paradigms of routing and matching brings revolutionary efficiency to ride-sharing creating a win-winsituation for all. We consider every technical factor, both real-time and historical, and perform a series of online and offline analytics to bring the best riding experience.

How we do it?

We consider every technical factor, both real-time and historical, and perform a series of online and offline analytics to bring the best riding experience.

Matching

Right from the moment when a rider selects pickup & drop locations to getting matched to a trip to eventually getting picked up and dropped off, the matching engine plays a crucial role in entire riding experience. Our matching engine optimizes the following factors:

• App Experience
• Wait time at the rider’s location

• Travel Times
• Quality of the ride

• Driver earnings per km driven
• Minimize driver idle time and wait time.

How do we do it ?

cab2share Matching Engine


  • ▪ Fast and configurable matching engine at the core allows us to experiment with various matching strategies while still maintaining a low latency.
    • ◦ The matching engine provides for an optimized way of calculating the cost of matching demand to a route and is configurable enough to choose various cost metrics like distance, time or a weighted function of the two.

  • ▪ In-memory processing allows for low latency operations.
    • ◦ Most of the processing and storage happens in memory and helps us execute matching requests at lightning speed while maintaining a stateless architecture for serving requests that helps us scale horizontally with load.

  • ▪ Exposed as a REST endpoint that allows us to scale easily with the load

Routing

cab2share is a one-stop solution to all commuting problems. Our diverse riding options offer full range of alternatives for varied commuting requirements. We believe making the right ride option available at the right location is the key to ensure the right balance between supply and demand which in turn guarantees economical prices, super easy riding experience and an efficient ride-sharing ecosystem.

How do we do it ?

cab2share Routing Engine


  • • Our static routing engine analyzes historical demand patterns, hourly traffic trends and rider SLAs to create optimal routes.
  • • High-speed dynamic routing engine ingests real time updates and performs feasibility checks at blazing speed to optimize riding experience.
  • • Machine learnt corrections improve the accuracy of our calculations and helps meet SLAs reliably.
  • • Ability to bootstrap from various sources of traffic allows us to hit the ground running in new cities.
  • • Running Simulations on historical trips helps determine riding patterns and frequently travelled paths and path segments to configure static trunk lines.
  • • Dynamic on-line fleet management around the trunk lines optimizes real-time demands.

Data Platform and Analytics infrastructure

Our cutting-edge routing and matching algorithms are built on top of massively scalable data platform and compute-intensive analytics infrastructure. The ability to capture millions of data points at the tick of a clock and run high-speed analytics on top of it empowers us to evaluate every technical factor in making the ride sharing efficient and friendly.

How do we do it ?

Logging and Analytics


  • 1. Integrated visualization to eyeball the state of the world and understand the network from a birds eye perspective.Many routing and matching issues stand out from a visual analysis of the data. Whether it is the the number of unfulfilled requests or SLA slippages, a birds eye perspective helps us overlay the data in a form that helps us catch the underlying issues.
  • 2. Real-time metrics for those metrics that cannot wait.It is essential that we monitor some metrics such as number of unfulfilled requests or SLA slippages to help ensure quality of service. The system provides for easy methods to pick and plumb such metrics for realtime monitoring.

A/B Testing


  • A/B testing capability enables us to test various strategy and choose the one that works best for a region.While a configurable matching engine helps to develop new matching strategies that improve customer experience, A/B testing helps us verify the impact in an unbiased and data-driven manner.

Use Cases