Executive Summary
This project consists of a data-focused approach to improve the livelihood of gig economy workers. While these workers often use their personal experience to optimize their earnings, these methods lack the power of the immense amount of data available in this technology-heavy industry. Our vision for Gig is a predictive algorithm that suggests strategies that maximize the utility of its users – to that end, it focuses on learning each driver's underlying valuation of costs associated with the gig economy such as miles driven and wait time, and make proposals accordingly.
Behind the scenes, the current implementation of the algorithm uses a multinomial logit discrete choice model which estimates the parameters that best explain the decision history for each driver. This model also requires estimates of the costs and rewards associated with the realized and counterfactual decisions; in order to obtain these values, we processed public and private historical data which included monetary, geospatial, and time components.
Currently, we have a working proof-of-concept, which we expect to further improve by developing a working user interface, expanding the current parameter estimation method, and implementing a crowdsourcing component.
Log in or sign up for Devpost to join the conversation.