Comment
Author: Admin | 2025-04-27
Uber is one of the most interesting companies in terms of the data science task complexity that needs to be done to run their businesses. Obviously the business is diverse having ride sharing, food delivery, autonomous mobility and possibly others. Here we will focus on describing some of the problems and how they have been accommodated through multiple generations of their in-house developed ML architecture. Problem statements # Financial Planning #Uber launched in San Francisco and this meant that they would be interacting regularly with the tech community who are continually looking for new tools and services that improve their quality of life. Uber took aim at those people by sponsoring tech events, providing free rides, and in general driving awareness among this audience. By seeding this audience, they were able to create a growth engine that hinged on the fact that these adopters would show their friends, who would become new users after their first Uber experience. Leading to a growing network of passionate customers ref.Uber needs to do budget allocation for thousands of markets (cities) across the globe of their seeding moneys. This resource allocation is implemented via financial planning backend services that generate and evaluate scenarios for each market. The solution involves a combination of optimization approaches such as the Stochastic Gradient Descent (SGD) that we will treat in a later chapter and the usage of the Michelangelo platform to clean data, train, and serve models that will create the necessary metrics for each scenario - these are the variables of the objective or constraints in the budget optimization.Financial planning per market. Squares are metric (aggregations) and ovals are computational elements (models). Suppose the seeding process gives the São Paulo city team 200 dollars for new user acquisitions (acquisition_spending metric) and 100 dollars for existing user retention (engagement_spending metric). The cost curve model tells us that setting 200 dollars for the acquisition_spending metric will get us 35 new rider signups. By running our models in topological order, we can compute our metrics, including the very last one, a net_inflow of 1,274 dollars. UberEATS #Estimated Delivery Time (EDT) is the most important prediction affecting customer experience in the delivery industry.Predicting meal estimated time of delivery (ETD) is not simple. When an UberEATS customer places an order it is sent to the restaurant for processing. The restaurant then needs to acknowledge the order and prepare the meal which will take time depending on the complexity of the order and how busy the restaurant is. When the meal is close to being ready, an Uber delivery-partner is dispatched to pick up the meal. Then, the delivery-partner needs to get to the restaurant, find parking, walk inside to get the food,
Add Comment