US 11,900,496 B2
Systems and methods for transport cancellation using data-driven models
Snir Kodesh, San Francisco, CA (US); Andrew Li, San Francisco, CA (US); Desmond Mawuko Torkornoo, Oakland, CA (US); and Naomi Yarin, Oakland, CA (US)
Assigned to Lyft, Inc., San Francisco, CA (US)
Filed by Lyft, Inc., San Francisco, CA (US)
Filed on Aug. 12, 2022, as Appl. No. 17/886,831.
Application 17/886,831 is a continuation of application No. 16/027,208, filed on Jul. 3, 2018, granted, now 11,449,962.
Prior Publication US 2023/0023453 A1, Jan. 26, 2023
Int. Cl. G06Q 50/30 (2012.01); G06N 20/00 (2019.01)
CPC G06Q 50/30 (2013.01) [G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
receiving, by a dynamic transportation matching system from a computing device of a requestor, a request for transportation specifying a pickup location for the requestor;
matching, by the dynamic transportation matching system, the requestor with a provider for completion of the request;
sending, by the dynamic transportation matching system to a computing device of the provider, the request for transportation;
calculating, by the dynamic transportation matching system, an estimated target for arrival of the provider at the pickup location based on an initial location of the provider;
monitoring, by the dynamic transportation matching system, a progress of the provider towards the pickup location, the monitoring including identifying a subsequent target for arrival of the provider at the pickup location based on a subsequent location of the provider;
calculating, by the dynamic transportation matching system, a probability of a cancellation of the matching of the requestor with the provider based on the progress of the provider towards the pickup location by:
determining that the pickup location is part of a geographic cluster;
generating a data-driven model using machine learning;
tuning the data-driven model for the geographic cluster; and
using the data-driven model associated with the geographic cluster to calculate the probability of the cancellation of the matching; and
cancelling, by the dynamic transportation matching system, the matching of the requestor with the provider based at least in part on:
accessing the data-driven model associated with the geographic cluster to determine a threshold value for the probability of the cancellation of the matching; and
comparing the probability of the cancellation of the matching of the requestor with the provider to the threshold value.