Developing a predictive algorithm for selecting matching candidates in dynamic carpool application

Rideflag是一家开发按需拼车系统的初创公司。使用RideFlag的移动应用程序，乘客可以要求拼车，RideFlag系统将乘客的路线与这些路线上可用的司机进行匹配，从而使双方都能节省资金并减少拥堵。在过去的12个月里，该系统一直在积极地进行拼车匹配。但是随着用户数量的增加，可能的路由匹配成倍增加，系统面临着可扩展性问题。这个合作项目的目标是开发一种新的算法，用于对可能的路由进行粗过滤，使系统能够扩展到一个大的用户和路由池。这将使系统在保持其性能的同时具有成本效益。我们建议使用先进的机器学习和大量数据来开发这种能够适应新环境的新算法。

Rideflag is a startup company developing a system for carpooling on demand. Using RideFlag's mobileapplication, riders can request a carpool and the RideFlag system matches riders routes with available driverson these routes enabling both parties to save money and reduce congestion. The system has been actively doingcarpool matches for the past 12 months. But the system is facing a scalability problem as the number of usersincrease which possible route matches increase exponentially.This collaborative project goal is to develop a new algorithm for gross filtering of possible routes to enable thesystem to scale up to a large pool of users and routes. This will allow the system to maintain its performancemetrics while being cost effective. We propose using advanced machine learning coupled with massive data todevelop this new algorithm that can adapt to new environments.