Real Life Machine Learning Case on Mobile Advertisement: A Set of Real-Life Machine Learning Problems and Solutions for Mobile Advertisement

摘要

This paper is an output of data science study on a real life problem. The paper starts with the problem definition and a brief introduction to the mobile advertisement for addressing the machine learning problems. Later on, some machine learning solutions are provided for each of the problems, furthermore the success of classical solution methods in the literature is also compared for the real life problems. Some problems addressed are: unbalanced data sets, parameter optimization, time slicing and history optimization and there are also some performance metrics related to the mobile advertisement problem domain. This paper mainly considers the actions generated by users and advertisement providers as a data stream and proposes a well optimized recommender algorithm based on crucial parameters. Different than most of the papers in the literature, this study is an output of a research collaboration with a real life advertisement platform.