Large scale ad click prediction system

Abstract

Ad click prediction is an important and challenging machine learning problem, both because it tries to find most related advertisements to visitor and because it increases the profitability of advertisers and the advertising platform. In this study, additional difficulties are given and solution proposals are presented when a frequently used model for ad click prediction is implemented to a large scale system. The detailed results obtained from this work show that the proposed solutions work well in terms of performance and success.