Browse a Model Using the Microsoft Sequence Cluster Viewer

Applies to: SQL Server 2019 and earlier Analysis Services Azure Analysis Services Fabric/Power BI Premium

The Microsoft Sequence Cluster Viewer in Microsoft SQL Server SQL Server Analysis Services displays mining models that are built with the Microsoft Sequence Clustering algorithm. The Microsoft Sequence Clustering algorithm is a sequence analysis algorithm for use in exploring data that contains events that can be linked by following paths, or sequences. For more information about this algorithm, see Microsoft Sequence Clustering Algorithm.

Viewer Tabs

When you browse a mining model in SQL Server Analysis Services, the model is displayed on the Mining Model Viewer tab of Data Mining Designer in the appropriate viewer for the model. The Microsoft Sequence Cluster Viewer provides the following tabs for use in exploring sequence clustering mining models:

• Cluster Diagram

• Cluster Profiles

• Cluster Characteristics

• Cluster Discrimination

• Cluster Transitions

Cluster Diagram

The Cluster Diagram tab of the Microsoft Sequence Cluster Viewer displays all the clusters that are in a mining model. The shading of the line that connects one cluster to another represents the strength of the similarity of the clusters. If the shading is light or nonexistent, the clusters are not very similar. As the line becomes darker, the similarity of the links becomes stronger. You can adjust how many lines the viewer shows by adjusting the slider to the right of the clusters. Lowering the slider shows only the strongest links.

By default, the shade represents the population of the cluster. By using the Shading****Variable and State options, you can select which attribute and state pair the shading represents. The darker the shading, the greater the attribute distribution is for a specific state. The distribution decreases as the shading gets lighter.

To rename a cluster, right-click its node and select Rename Cluster. The new name is persisted to the server.

To copy the visible section of the diagram to the Clipboard, click Copy Graph View. To copy the complete diagram, click Copy Entire Graph. You can also zoom in and out by using Zoom In and Zoom Out, or you can fit the diagram to the screen by using Scale Diagram to Fit in Window.

Back to Top

Cluster Profiles

The Cluster Profile tab provides an overall view of the clusters that the algorithm in your model creates. Each column that follows the Population column in the grid represents a cluster that the model discovered. The <attribute>.samples row represents different sequences of data that exist in the cluster, and the <attribute> row describes all the items that the cluster contains and their overall distribution.

The Histogram bars option controls the number of bars that are visible in the histogram. If more bars exist than you choose to display, the bars of highest importance are retained, and the remaining bars are grouped together into a gray bucket.

You can change the default names of the clusters, to make the names more descriptive. Rename a cluster by right-clicking its column heading and selecting Rename cluster. You can hide clusters by selecting Hide column, and you can also drag columns to reorder them in the viewer.

To open a window that provides a larger, more detailed view of the clusters, double-click either a cell in the States column or a histogram in the viewer.

Back to Top

Cluster Characteristics

To use the Cluster Characteristics tab, select a cluster from the Cluster list. After you select a cluster, you can examine the characteristics that make up that specific cluster. The attributes that the cluster contains are listed in the Variables columns, and the state of the listed attribute is listed in the Values column. Attribute states are listed in order of importance, described by the probability that they will appear in the cluster. The probability is shown in the Probability column.

Back to Top

Cluster Discrimination

You can use the Cluster Discrimination tab to compare attributes between two clusters, to determine how items in a sequence favor one cluster over another. Use the Cluster 1 and Cluster 2 lists to select the clusters to compare. The viewer determines the most important differences between the clusters, and displays the attribute states that are associated with the differences, in order of importance. A bar to the right of the attribute shows which cluster the state favors, and the size of the bar shows how strongly the state favors the cluster.

Back to Top

Cluster Transitions

By selecting a cluster on the Cluster Transitions tab, you can browse the transitions between sequence states in the selected cluster. Each node in the viewer represents a state of the sequence column. An arrow represents a transition between two states and the probability that is associated with the transition. If a transition returns back to the originating node, an arrow can point back to the originating node.

An arrow that originates from a dot represents the probability that the node is the start of a sequence. An ending edge that leads to a null represents the probability that the node is the end of the sequence.

You can filter the edge of the nodes by using the slider at the left of the tab.

Back to Top

See Also

Mining Model Viewer Tasks and How-tos
Mining Model Viewer Tasks and How-tos
Microsoft Sequence Clustering Algorithm
Data Mining Tools
Data Mining Model Viewers
Browse a Model Using the Microsoft Cluster Viewer