Viewing a Mining Model with the Microsoft Tree Viewer
The Microsoft Tree Viewer in Microsoft SQL Server 2005 Analysis Services (SSAS) displays decision trees that are built with the Microsoft Decision Trees algorithm. The Microsoft Decision Trees algorithm is a classification and regression algorithm for use in predictive modeling of both discrete and continuous attributes. For more information about this algorithm, see Microsoft Decision Trees Algorithm.
When you build a decision tree model, Analysis Services builds a separate tree for each predictable attribute. You can view an individual tree by selecting it from the Tree list on the Decision Tree tab of the viewer.
A decision tree is composed of a series of splits, with the most important split, as determined by the algorithm, at the left of the viewer in the All node. Additional splits occur to the right. The split in the All node is most important because it contains the strongest split-causing conditional in the dataset, and therefore it caused the first split.
You can expand or collapse individual nodes in the tree to show or hide the splits that occur after each node. You can also use the options on the Decision Tree tab to affect how the tree is displayed. Use the Show Level slider to adjust the number of levels that are shown in the tree. Use Default Expansion to set the default number of levels that are displayed for all trees in the model.
Predicting Discrete Attributes
When a tree is built with a discrete predictable attribute, the viewer displays the following on each node in the tree:
The condition that caused the split.
A histogram that represents the distribution of the states of the predictable attribute, ordered by popularity.
You can use the Histogram option to change the number of states that appear in the histograms in the tree. This is useful if the predictable attribute has many states. The states appear in a histogram in order of popularity from left to right; if the number of states that you choose to display is fewer than the total number of states in the attribute, the least popular states are displayed collectively in gray. To see the exact count for each state for a node, pause the pointer over the node to view an InfoTip, or select the node to view its details in the Mining Legend.
The background color of each node represents the concentration of cases of the particular attribute state that you select by using the Background option. You can use this option to highlight nodes that contain a particular target in which you are interested.
Predicting Continuous Attributes
When a tree is built with a continuous predictable attribute, the viewer displays a diamond chart, instead of a histogram, for each node in the tree. The diamond chart has a line that represents the range of the attribute. The diamond is located at the mean for the node, and the width of the diamond represents the variance of the attribute at that node. A thinner diamond indicates that the node can create a more accurate prediction. The viewer also displays the regression equation, which is used to determine the split in the node.
Additional Decision Tree Display Options
When drill through is enabled for a decision tree model, you can access the training cases that support a node by right-clicking the node in the tree and selecting Drill Through. You can enable drill through within the Data Mining Wizard, or by adjusting the drill through property on the mining model in the Mining Models tab.
You can use the zoom options on the Decision Tree tab to zoom in or out of a tree, or use Size to Fit to fit the whole model in the viewer screen. If a tree is too large to be sized to fit the screen, you can use the Navigation option to navigate through the tree. Clicking Navigation opens a separate navigation window that you can use to select sections of the model to display.
You can also copy the tree view image to the Clipboard, so that you can paste it into documents or into image manipulation software. Use Copy Graph View to copy only the section of the tree that is visible in the viewer, or use Copy Entire Graph to copy all the expanded nodes in the tree.
The Dependency Network displays the dependencies between the input attributes and the predictable attributes in the model. The slider at the left of the viewer acts as a filter that is tied to the strengths of the dependencies. If you lower the slider, only the strongest links are shown in the viewer.
When you select a node, the viewer highlights the dependencies that are specific to the node. For example, if you choose a predictable node, the viewer also highlights each node that helps predict the predictable node.
If the viewer contains numerous nodes, you can search for specific nodes by using the Find Node button. Clicking Find Node opens the Find Node dialog box, in which you can use a filter to search for and select specific nodes.
The legend at the bottom of the viewer links color codes to the type of dependency in the graph. For example, when you select a predictable node, the predictable node is shaded turquoise, and the nodes that predict the selected node are shaded orange.
The Mining Legend displays the following information when you select a node in the decision tree model:
The number of cases in the node, broken down by the states of the predictable attribute.
The probability of each case of the predictable attribute for the node.
A histogram that includes a count for each state of the predictable attribute.
The conditions that are required to reach a specific node, also known as the node path.
You can dock and work with the Mining Legend in a similar manner as with Solution Explorer.