This selection displays a successfully built O-Cluster model (a clustering model).
The following menus appear when the model is displayed in a new window:
The model viewer the following tabs:
This tab displays the Type of the model (Clustering), the Algorithm used to build the model (O-Cluster), the Algorithm Settings used to build the model, and the Attributes used to build the model.
Oracle Data Mining 11g models can be built using Automatic Data Preparation (ADP). If the model was built using ADP, then Automatic Data Preparation has the value yes
.
The following settings are displayed:
For information about the settings, see O-Cluster Settings.
The attributes used to build the model are listed in the Attributes grid; the following information is displayed for each attribute:
NUMBER
, FLOAT
, CHAR
, VARCHAR2
, etc.To save the information about the attributes as a text file, click the Save As icon (a yellow arrow above a graphic of a page).
To see the clusters, click the Clusters tab.
Data for an O-Cluster model build is discretized (binned). If the bin table is available, Bin is available. Click Bin to use bin values in the rules; click Unbin to use unbinned values. The default is to use unbinned values.
The following constants are displayed:
You can display either all clusters (the default) or leaf clusters only. If you display all clusters, you can see how and why the splits were created in the iterative process. To see the final clusters, click the Show Leaves Only checkbox.
For each cluster, the Cluster ID, number of Cases, and Split Rule are displayed. (No split rule is displayed for a leaf cluster, because a leaf cluster is not split.)
To see the cluster details, including attribute histograms, select a cluster, and click the Detail button. For more about the detail display, see Cluster Details.
You can expand and collapse all cluster nodes by clicking the appropriate buttons.
To see results, click the Results tab.
The tab contains a tree view of results. To view specific results, select the result, and click View. To delete a result, select one or more results and click Delete.
To see the rules, click the Rules tab. To see the rules defined by a cluster, click the cluster.
You can restrict the rules displayed to those that involve the top-most relevant attributes; to do this, click the checkbox Show topmost relevant attributes. The default number of attributes is 10; you can change this value. Click Refresh to apply the restriction. The default is to not restrict rules in this way.
Data for an O-Cluster model build is discretized (binned). If the bin table is available, Bin is available. Click Bin to use bin values in the rules; click Unbin to use unbinned values. The default is to use unbinned values.
You can display either rules for all clusters (the default) or rules for leaf clusters only. To display rules for leaf clusters only, click the Only Show Rules For Leaf Clusters checkbox. You can also reduce the complexity of rules by limiting the attributes included to the most important attributes. To limit the attributes included in rules, click the Only Show Attributes with Minimum Relevance Rank checkbox, and specify the minimum relevance rank. The default minimum relevance rank is 10. For a description of relevance rank, see Relevance Rank. After you change either of these items, click Refresh to display the rules that satisfy the specified requirements.
The rules are displayed in a grid.
To sort the rules, click Sort and specify how you wish to sort the rules. The default is to sort by descending order of rule confidence.
To view a detailed display of a rule, select the rule; the detailed rule is displayed in the Rule detail area. For each rule, a statement of the form IF followed by a condition and THEN followed by cluster number equals number (number is the cluster number) is displayed.
The following is displayed for each rule:
Support and confidence can be used to rank the rules and hence the predictions.
This tab describes the task used to build the model. It shows when the task ran, what input was used, and what output was created.
Copyright © 2006, 2008, Oracle. All rights reserved.