Data Mining algorithms accept training data from relational or OLAP data sources. When you mine OLAP cubes, your case table is a dimension. Any measure from any measure group (fact table) connected with the selected dimension can be used as a case-level column. Additionally, you can add other dimensions connected to the same measure group as nested tables.
I have setup a virtual machine on Hiper-V to analyze the SQL Server Denali CTP3 bits. One of the “new” features I was interested on was the upgrade on the SSAS side. As probably most of you have been reading, there is a double path on the Analysis Services area, to go on the tabular or the OLAP model. To solve that it has been defined the BISM (Business Intelligence Semantic Model), you can find additional information ‘La biblioteca de PowerPivot’ from my SolidQ Colleagues. What I have found has really surprised me because you have to decide on the installation path (as you can see in the figure below) if you want to setup a tabular or a multidimensional instance. That means that the SSAS engine is totally different and you have to setup separate instances names even if you want to have both models in your company. (more…)