Grab

How We Harnessed the Wisdom of Crowds to Improve Restaurant Location Accuracy

Apr 2, 2019Pravin Kakar
How We Harnessed the Wisdom of Crowds to Improve Restaurant Location Accuracy
SummaryAI, from the source post

This post details the methodology for inferring restaurant pickup locations using driver GPS data and timestamps. It explains the filtering of sub-par GPS data, the definition of 'at the restaurant' based on time spent within a radius and longest streak analysis, and the ranking criteria (order volume, fraction of invalid pickups, median distance) used to prioritize restaurants for manual verification. The results show a reduction in median error distance and a 5x decrease in cancellations due to incorrect locations.

More from the Restaurant Location Accuracy Improvement trail