Cluster Profiling

To explore clusters, use the DM Example-1 branch, which starts with the DM Example-1 node and ends with the JMP Data Exploration node. The JMP Data Exploration node explores clusters using the JMP Graph Builder and the mapping capabilities provided by JMP. Follow the steps below to examine the results of the DM Example-1 branch.

Select the JMP Data Exploration node. Right-click the node and select Run. In the Confirmation window, select Yes.
After the process flow diagram has successfully run, select Results in the Run Status window.
In the Results window, click the View button.
By default, the target variable is plotted in the Graph Builder. In this example, the target variable is PURCHASE.
To create a segment plot, first click the Start Over button in the upper left corner of the Graph Builder window.
In the Variables list, select _Segment_ and drag it to the Y-axis. Next, select the variable INCOME and drag it to the X-axis. You should see a cluster profile plot that is similar to the one below.

Alternatively, if you select the Overlay option in the upper right corner of the Graph Builder window, the _SEGMENT_ variable is used as an overlay variable.

Both versions of the chart suggest that income is not a strong differentiator of the clusters. Segments 1 and 3 have similar interquartile ranges. Segment 2 has a higher median than the other segments, but it represents only one customer.

The Home Furniture variable (HOMEACC) is a better differentiator of segments 1 and 3. Segment 1 has a median of 6 home accessories, whereas segment 3 has a median of 0.