Mirror of Oracle documentation
Converted for search and offline reading. Authoritative source: Oracle. Diagrams and some complex tables are simplified — check the PDF when in doubt.
2 Customer Decision Trees
This chapter describes the use of the Customer Decision Tree Science Cloud Service module.
Introduction
A customer decision tree is a decision support tool that uses a tree-like graph to model a customer hierarchical decision-making process for a specific product. The branches of the tree provide a visual representation of the choices a customer makes and the order of importance of various product characteristics. Transaction data is used in the analysis.
A customer decision tree identifies the decisions a customer makes when choosing a particular product. The decision tree is produced by algorithms that analyze historical customer sales data. It illustrates how customers shop and how they evaluate the importance of different product attributes when making buying decisions. Such information can be useful to a retailer in terms of product selection and display.
The CDT Cloud Service module consists of three tabs: Overview, Generate CDTs, and Manage CDTs. You use the Overview tab to keep track of the status of each stage within the Generate CDTs tab. You use the Manage CDTs tab to assess the set of CDTs you and other users have created using the Generate CDTs tab, setting versions as complete, approving versions, and deleting versions, as necessary.
Overview of CDT Process
When you use the CDT Cloud Service module, you follow this general iterative process to create and manage CDTs:
-
Setup. You define the domain of calculation by selecting the categories and time frame for the calculation of the CDTs.
-
Data Filtering. You configure filters that remove input data that might cause errors in the calculation or that might lead to inaccurate or unreliable answers.
-
Calculation. You calculate a specific version of a CDT. Different versions of the CDT based on different configurations can be calculated and compared.
-
Evaluation. You examine the results of the calculation and determine the reliability and accuracy of the answers. You can prune inaccurate or unreliable results.
-
Escalation. When you are satisfied with the results, you can set the escalation path to fill in the holes for partitions whose CDTs were removed during pruning by setting up a search path through the segment hierarchy and the location hierarchy. Then, you can set a version of a CDT as complete.
Within each stage, you set the required parameters. In some stages, once you have done that you click Run to initiate the process. To determine when a run is complete, go to the Overview tab and click the Refresh icon to see the current status of each stage. If you make changes to a stage, you must re-run that stage and all the relevant stages that follow it as the results of those stages are made invalid by the changes you made.
ee
Table 2-1 (Cont.) CDT Generation Stage Status Fields
| Field Name | Description |
|---|---|
| Last Update Date | The date and time when the status was most recently updated. |
The following provides an explanation of the different values for the status of a stage.
Table 2-2 Status Explanations
| Stage Status | Description |
|---|---|
| Not Started | This indicates that the stage has not yet been initiated. This can occur when you first begin to create a CDT or when you update an existing version. |
| Not Started (Scheduled for Later) | The stage is scheduled to start after the previous stage is complete. |
| Processing | This indicates that the stage is currently being processed. |
| Stopped/Paused | This indicates either that the stage has been stopped by the application or that the user has chosen to stop the process (for example, to make a change to an option). In the latter case, once any background processing has stopped, the user can re-run the stopped stage. |
| Completed Successfully | This indicates that the stage has been successfully processed. |
| Completed with Errors | This indicates that the stage was able to complete the processing of all requested CDTs, but one or more CDTs encountered errors during the run and were not able to complete successfully. This most commonly occurs when the data used for a CDT is not available or is too sparse to produce a result. |
| Failed | This indicates that a problem occurred during the processing. If the failure occurred during the Calculation stage, and only a few CDT results did not produce a CDT score, then the user can continue without those results since pruning will remove them. If the failure occurred during any of the other stages, the user must re-run the stage. |
| Cancelled | This indicates that the database has cancelled the execution of the stage either because of missing data or an exception. The user should review the RSE_LOG_MSG log to determine the problem. |
Recent Run Details
The Recent Run Details displays the CDT results for the indicated version from the last successful run, if one has occurred.
Here you can review information by Categories, Location, or by Customer Segment by using the filter. You can also set a version as complete, view and edit a CDT, or compare two CDTs.
Click the Refresh icon to update the fields and see the latest information in this table.
ee
ee
Set Version as Complete
Each partition must have an active CDT. The active CDT is used by other applications that require a CDT.
All CDTs within a version are activated when you click Set Version as Complete .
If a CDT exists in more than one version, then the most recently activated version takes precedence.
If a version contains multiple categories and is Complete and then a different version with a partial overlap of categories is later marked as Complete, only the overlapping categories are replaced in the new version.
If a version is overwritten, then the CDTs in that version are no longer active.
When a CDT is at risk of going from active to inactive, you will see a warning message.
To determine whether or not a CDT is active, go to the Manage CDTs tab. The Browse by Categories or by Versions table displays a flag that indicates whether or not a CDT is active.
View One CDT
This button provides access to the CDT Editor. For details about this functionality, see Using the CDT Editor.
Compare Two CDTs
This button provides access to CDT Compare. For details about this functionality, see Comparing Two CDTs.
Approve a Version
Only one approved CDT is permitted for a given CMPO node/segment/location combination. If you approve a second CDT with the same combination, it overwrites the first one.
If an Approved CDT exists, then the Approved CDT is returned to the requesting application.
If no Approved CDT exists, then the Active CDT is returned to the requesting application.
If no Approved CDT exists and no Active CDT exists, an error is returned to the requesting application.
Approval can be done for individual CDTs or for an entire version; activation is for an entire version.
An approved version is used by the CMPO application. An active version is used by Demand Transference.
Unapprove
Use this to change the status of a CDT that was previously approved. Select a single CDT record in the data table and click the Unapprove button to remove it from the approved version.
Delete a Version
If you delete a version, it cannot be activated. You cannot delete an approved version, but you can remove a CDT from an approved version.
|
Add a Child
To add a child node to the CDT, do the following:
1. Click Add Child . You see the Add CDT Node window.
2. Select an attribute from the Attribute drop-down menu.
3. Select a value from the Attribute Value drop-down menu.
4. Select the Create separate branch per value check box, if applicable. When more than one of the attribute values are selected, this field create a separate node for every attribute value selected.
Note
A node can be created by selecting one or more attribute values. If multiple attribute values are chosen, and the check box to create a separate branch is not selected, one node that represents multiple attribute values is created.
If one branch is created for all values of an attribute, select all attribute values and only one node will appear with the attribute value of ALL.
If all the attribute values for an attribute are used, either by creating one node or separate nodes, a child cannot be added to the node and the Add Child is disabled. Creating a node with ALL does not associate the specific attribute values with the node. Instead, it encompasses any attribute value.
5. Click OK .
Editing Nodes
Editing is allowed on all nodes as long as there are additional attribute values available. The selections from the node being edited are preselected in the dialog. The attribute cannot be changed, so the attribute drop-down menu is not disabled. Any attribute values that are in use by other siblings are not be displayed in the list.
Deleting Nodes
Nodes can be deleted from the tree. There are two options for deleting: Delete and Delete Branch.
Selecting Delete deletes the selected node. Delete is available for any leaf node (a node without any children). It is also available for a non-leaf node that represents all the attributes for an attribute value: the children under this node move under the parent.
Selecting Delete Branch deletes the entire branch of nodes under the selected node, but not the selected node itself.
Copying/Pasting in Customer Decision Tree Editor
Copy and Paste functionality is provided to copy nodes from one branch to another. Copy Branch is enabled when a node is selected that has children. The Copy Branch function copies the full branch of children of the selected node, but not the selected node itself. Paste Branch is enabled when a branch has been copied and the selected node is a leaf node.
ee
ieSSSSCisN
eee
ee
ee
ee
SCC‘
Table 2-20 (Cont.) Escalation Report
| Field Name | Description |
|---|---|
| Location Level | Identifies the location level. |
| Number of CDTs | The number of partitions that are trying to have a CDT assigned. This is generally the number of customer segments by the number of locations. |
| Percentages | The percentage of partitions that have been assigned a CDT from a given escalation level. |
Completion of Process
When a version is complete, the results for the version are activated so that other applications can use the information. The similarity data that has been calculated during the generation process is also activated for use.
After the completion of this step, the intermediate results from each stage are removed from the database and can no longer be used.
In this guide
- Guide: AI Foundation User Guide
- Previous: 18 Appendix: Forecast Approval Exceptions
- Next: 3 Demand Transference