For small and medium sized internet-based companies, such as online gaming publishers, online advertising has become an important business model. Many publishers are forced to use existing SSPs (Google, LiveRail, OpenX, Appnexus, Smaato, Rubicon, etc.) that operate with different pricing models and have a heterogeneous advertiser portfolio. PADS is aimed to improve the practice significantly by making systematic analysis of which advertisement slots are more profitable with machine learning techniques and to model the SSP selection as a resource harvesting problem with the learned models. Systematic analysis of which slots are more robust in which SSPs using machine learning techniques, modeling SSP as a resource allocation problem, and optimizing layer configurations for selling different AD slot groups based on learned prediction models are being studied in PADS.

C#, .NET Framework 4.5.2, ASP.NET MVC, Entity Framework 5.0, ADO .Net, Console Application, Object-Oriented Programming, Javascript, SQL Server 2014, Machine Learning, Data Mining, SVMs, ANNs, Decision Trees, Density Estimation, Stochastic Optimization