Davorin Strehar (2014) Optimizing number of online ad impressions per user. EngD thesis.
Abstract
This thesis deals with the optimization of displaying online ads (frequency capping), where ad serving represents a cost, and clicks on the ads represent advertising revenue. In this case we would like to display the ad to the user so many times to have the biggest probability of the ad click. For this purpose, we converted the ad serving data to the appropriate format, analyzed it and used machine learning methods to make predictive models with the aim of predicting the optimal number of viewable impressions per user on a specific ad campaign. Afterwards, we used these predictive models in the ad serving simulation where we measured the performance of each procedure. The result of the thesis is the selection of the best procedure to assess the optimal number of viewable ad impressions per user.
Actions (login required)