Databases Reference
In-Depth Information
to Predict. The value Predict means that the Bike Buyer attribute will
be used as an input as well as an output to the mining model. By
choosing the Bike Buyer as an input column you are providing addi-
tional information to the mining model algorithm so that the model
accurately represents the input data.
16. Having completed all the refinements to the decision tree model, you
can now deploy this model to the server similar to the cube or dimen-
sion. Hit the F5 button to deploy the mining model.
Business Intelligence Development Studio sends the definition of the entire
project you created to the server along with a process request. Once the
database is processed, the BIDS switches the view to the Mining Model view-
er as shown in
Figure 14-12
. The decision trees algorithm identifies the
factors influencing customers to buy bikes and splits customers based on
those factors and stores it within the model. The information stored in the
model is better visualized using a tree structure. The mining model viewer
represents the contents of the mining model in the form of a tree view which
contains a series of nodes. The root of the tree starts with a single node that
represents all the customers. Each node shows the percentage of customers
(shown by the horizontal bar within the node) who have bought bikes based
on the input set. Each node is split into multiple nodes based on the next
most important factor that determines why a customer has bought a bike. The
tree contains nodes at several levels from 1 to N based on the number of
splits determined by the decision tree algorithm. Each node in the tree is as-
sociated to a specific level in the tree. The root node is at level 1 which is the
top most level. The depth of the tree is measured by the number of splits or
levels of the tree. In the mining model viewer you can make the selection to
view the tree up to a specific level by selecting the option from Default Expan-
sion or Show Level.
Figure 14-12
shows nodes with the horizontal bar that is
shaded with two colors. When you are using the product you will see these
shaded regions of the horizontal bar in colors red, blue, or both. The mining
legend window shows the legend for the colors in the horizontal bar or a
node. If the legend window is not visible, right-click in the mining model view-
er and select Show Legend.
Search WWH ::
Custom Search