|
Briefing
|
- Ride Matching Algorithm – Ride hailing firm Didi Chuxing testing new algorithm for assigning drivers to rides in select cities
- Reinforcement Learning – Uses reinforcement learning, form of machine learning that accepts rewards and penalties for accomplishing goals
- More Integrated – Combines forecasting system which predicts changes in rider demand and matching system that assigns drivers to rides into one algorithm, instead of having two parts in old algorithm
- Simplified Process – Integrated approach removes need to retrain new algorithm if patterns of driver supply and rider demand change, unlike in old algorithm which cannot evolve with changing supply and demand
- Greater Efficiency – A/B Testing found new algorithm is more efficient than older algorithm
- Initial Rollout – Plans rollout in select Chinese cities, with timeline undisclosed
|
|
Accelerator
|
|
|
Sector
|
Information Technology, Transportation and Logistics
|
|
Organization
|
DiDi Chuxing
|
|
Source
|
|
|
Original Publication Date
|
December 12, 2018
|
