Explore a data-driven matching algorithm for ride pooling in this 32-minute DS4DM Coffee Talk presented by Ismail Sevim from Université de Montréal. Delve into the proposed machine learning algorithm based on rank aggregation, designed to match drivers and riders efficiently. Examine how the algorithm learns feature weights from historical data through an optimization model, and understand the transformation of the resulting nonlinear bilevel optimization model into a single-level mixed-integer nonlinear optimization model. Discover the algorithm's performance as demonstrated using a real-life dataset from a ride pooling start-up company's mobile application, with the company's current approach serving as a benchmark.
Overview
Syllabus
A Data-Driven Matching Algorithm For Ride Pooling Problem, Ismail Sevim
Taught by
GERAD Research Center