You can use the Slice Source Cube dialog box to restrict the data used to train the model. Typically a cube contains data related to many different dimensions and attributes, such as all stores, all regions, and all products. It is not practical to train a model on unlimited combinations of attributes, so you use this dialog box to choose a specific set to use in training a model.
For example, in the AdventureWorks cube, you might slice by a particular product line, region, or year, to get a portion of the data.
Select the hierarchy that you want to slice. You can choose any level of a hierarchy, but the attributes used in the model will be only at the level that you choose.
For example, if you choose the Geography hierarchy, and select Country as the level, you cannot build a model that uses City as the attributes. Conversely, if you choose City as the level of the hierarchy on which to slice, you cannot create a mining model based on Country.
Select the operator to use in building a slice expression.
For example, if you chose Geography as the hierarchy, you could select the operator = and then type “Europe” as the filter, to get cube data for Europe only.
Type an expression to use as a criterion when filtering the cube on the selected dimension.