Lyft, Inc.

Christopher Sholley

Ricky Chachra

Adriel Frederick

Ashivni Shekhawat

Tzu-Hsin Chiao

Jerome Hong-Phat Thai

Jatin Chopra

Jianzhe Luo

Adam Greenhall

Chinmoy Dutta

Garrett Van Ryzin

Jimmy Young

Jeremy Alexander Karp

Danial Afzal

This disclosure covers machine-learning methods, non-transitory computer readable media, and systems that generate a multiplier that efficiently and effectively provides on-demand transportation services for a geographic area. The methods, non-transitory computer readable media, and systems dynamically adjust the multiplier with machine learners to maintain a target estimated time of arrival for a provider device to fulfill a request received from a requestor device. In some embodiments, the methods, non-transitory computer readable media, and systems generate a multiplier report comprising a representation of a geographic area and an indication of the multiplier to facilitate inflow and outflow of provider devices within and without the geographic area.

Dynamically generating and updating multipliers for a transportation matching system using machine learning

Embodiments provide techniques, including systems and methods, for assignment and/or reassignment of transport requests received within a request matching time period. For example, transport requests received during a request matching time period, as well as requests determined to be eligible during the request matching time period for reassignment, are associated with a location identifier (e.g., geohash) and are pooled at a dynamic transportation matching system in order for a dynamic assignment and/or reassignment that may reduce a metric (e.g., an overall estimated time of arrival (ETA) for requests) associated with the dynamic transportation matching system.

Dynamic optimized reassignment of providers at a geohash level

The disclosed computer-implemented method may include matching transportation requests to transportation providers using a dynamic transportation matching system. Matching transportation requestors which have varying preferences for waiting times and transportation pricing with transportation providers having varying levels of available supply may involve matching methods that satisfy the transportation requestors while maintaining an appropriate level of supply. A method which provides immediate service for transportation requestors that may be willing to pay a premium fare while placing transportation requestors not willing to pay a premium fare and may be willing to wait may provide a balance of supply and demand of transportation providers. Accordingly, immediately matching transportation requestors not willing to wait while placing the transportation requestors willing to wait in a queue may provide the balanced approach to managing transportation provider availability. Other methods, systems, and computer-readable media are disclosed.

Systems and methods for queueing in dynamic transportation networks

The present disclosure relates assigning ride requests to providers based on the probability that a provider will accept the request. For example, one or more embodiments identify a first provider to assign the ride request. The system then generates a probability of acceptance for that provider. The system then determines an estimated time-to-arrival for the first provider and an alternate estimated time-to-arrival associated with a re-assigned provider. Based on the acceptance probability, the estimated time-to-arrival, and the alternate estimated time-to-arrival, the system then determines an expected time-to-arrival associated with the first provider. The system assigns ride requests to providers based on optimizing the expected time-to-arrival across multiple potential providers for a ride request.

Assigning rides based on probability of provider acceptance

Growing community of inventors

San Francisco, CA, United States of America

Christopher Sholley