JT on EDM, James Taylor, Sep 8, 2010
I got an update from Microsoft recently. We covered lots of different products in and around data/analytics - some public, some under NDA. Microsoft is making a serious investment in SQL Server as well as into SharePoint and Excel. BI tools, they say, are not getting to most people and Microsoft sees this as an opportunity. They are targeting business users, bright domain experts who know where to find data and how to massage them in Excel. They estimate that there are some 500M such information workers of which about 100M could be considered power BI users even though not using a formal BI tool. Microsoft is targeting these folks with extensions to Excel, with PowerPivot (described as a way to mashup data in Excel) and with Report Builder 3.0 (making it easier to build operational reports). Extra use of data will mean increased IT demand to support them, so Microsoft is making a parallel investment in IT tools to see how taxing the queries are, which data sources are being used etc.
Microsoft takes a quite different perspective when it comes to Predictive Analytics. Traditionally products for predictive analytics are targeted at very high end users - statisticians, mathematicians. These users build sophisticated models that then must be interpreted for their users (executives, systems). Microsoft's plan was to bring modeling to a broader audience by targeting analysts without a lot of data mining or statistics expertise/experience. They also wanted to target developers without a huge background in statistics as well as people who needed to do some data mining in their day to day job. The result is a data mining stack on top of SQL Server with Analysis services sitting above Integration Services and below Reporting. Analysis Services supports OLAP, PowerPivot and data mining. The data mining environment provides an extensive editing environment and 9 popular algorithms. It is designed to be easy to integrate with Office and the rest of SQL Server as well as easy to embed through SQL based API. Finally a set of APIS make it extensible so companies can add additional algorithms.