Architecture of Predictive Programming
Predictive Programming harnesses the power of Data Mining (component of Microsoft SQL Server 2008 Analysis Services) to bring a form of artificial intelligence to your applications. It can be used for many purposes, but we concentrate on a way to place a negative-feedback loop into your software that helps it make better decisions. Specifically, we improve business process success rates by spotting problems and flagging them before a transaction completes. For instance, we can predict that an online purchase transaction is going to fail and we can flag it for attention so saving a customer frustration while protecting your reputation. Similarly, we can use this technique to perform predictive input validation. This session discusses the overall architecture of such applications, briefly introduces the use of Data Mining, and shows you examples of code that implements this approach. We only briefly discuss the Data Mining techniques used such as neural networks, decision trees, and clustering.