Web Economics is a graduate-level CS module I pursued while studying abroad at University College London, in the Spring of 2017. Its coursework component (30%) consisted of the following team project, whose abstract I present below. The project included literature review; dataset exploration and review; training a classifier and devising linear and non-linear pricing strategies to maximize CTR (Click-Through Rate) of online ads; evaluating our solution in terms of metrics such as Cost per Click, Conversion Rate, and total cost.

Real-Time Bidding (RTB) is an increasingly popular approach to online advertising that has evolved into a multi-billion dollar industry. An efficient approach to quickly and accurately optimizing an advertiser’s bidding strategy in ad auctions is paramount to the success of the advertising party, as a good solution can save vast amounts of money and lead to higher conversion rates than competitors.

In this paper, different approaches to automating the generation of bid prices for online users in real time are undertaken, and each is evaluated using common evaluation metrics, as well as comparatively with others. First, two naive strategies with little optimization are implemented, and results are presented and discussed. Subsequently, two machine learning approaches are presented, one using a linear model and one using a non-linear one. Our approach and steps in building our machine learning classier and developing a bidding strategy is detailed. Finally, the results of our best bidding strategy are presented and commented on using relevant evaluation metrics.

Project Report: The full project report can be found here.

Github link: Real-time bidding