OLAP Miner searches through an area of a cube that you define. The area of the cube that you define is called the subcube. The subcube structure is specified in a deviation detection definition. The deviation detection definition specifies which members from the hierarchy of each dimension that you want OLAP Miner to mine for deviations.
To create the deviation detection definition, you follow these general steps:
Specify a name for the deviation detection definition.
Figure 5. Deviation Detection wizard (1 of 3)
In this step, you create a deviation detection definition by defining an area of the OLAP cube (Demo Basic) that you want OLAP Miner to search.
OLAP Miner will search for deviations specifically in actual sales but in the context of the other dimensions.
Figure 6. Deviation Detection wizard (2 of 3)
OLAP Miner displays the selections that you made in the deviation detection definition. You can specify how many deviations that you want OLAP Miner to return. Leave the default number at 100.
Review the selections. Then click Save and run now.
Figure 7. Deviation Detection wizard (3 of 3)
OLAP Miner displays a Progress Display window to show how the mining run is progressing. Because you are using a relatively small cube in this example, the mining run should take only a few minutes.
Close the Progress Display window.
In this step, you view the mining run results.
Figure 9. OLAP Miner main window
You can sort the list of deviations by clicking on the heading of a column in the main window.
The first five columns represent the dimensions that you selected in the deviation detection definition. The Expected column describes whether the calculated expected value is higher or lower than the actual value of the cell. OLAP Miner uses the difference between the expected value and the actual value to calculate magnitude. As the difference between the expected value and the actual value increases, the magnitude increases. Higher magnitude values show greater disparities in the data.
Figure 10. Deviation Viewer, Product by Market view
The deviation that you selected is shown in red with a red border. Other deviations in this area of the subcube are highlighted in red.
Notice that the deviation -274 is significantly different from the other surrounding values. The actual profits in August for televisions in New York is much lower than expected. OLAP Miner calculates the expected value based on all the other values for each dimension that intersects with the cell Television, New York, August, Actual, Profit.
In this example, OLAP Miner determines that the Product by Market view is the most significant. The tabs under the deviations are shown in order of importance with the most important view shown first.
For more information about how OLAP Miner calculates expected values and magnitudes, see The OLAP Miner algorithm.
Figure 11. Deviation Viewer, Market by Year view
When you hold the cursor over a red highlighted value, the magnitude and expected values are displayed.
If you installed the DB2 OLAP Server spreadsheet add-in: