Data Mining at Microsoft

David Heckerman

For the last eight years, I have been working with a team at Microsoft to develop principles, methods, and technology for data mining. Some of this technology now appears in Micrsoft products including SQL Server 2000 and Commerce Server 2000. Our research has covered several important aspects of data mining including exploratory data analysis, large-scale data analysis, automated data analysis, causal discovery, and the integration of data analysis and data storage. Several classes of statistical models including graphical models and mixtures models have played an important role. In my talk, I will describe our research and demonstrate our technology using real examples from web mining and ecommerce.