Visual Six Sigma: Making Data Analysis Lean

4.4. Uncovering the Hot Xs

At this point, Alex and Alice have learned a great deal about their data. They are ready to begin exploring the drivers of late charges, to the extent possible with their limited data. They are interested in whether late charges are associated with particular accounts, charge codes, or charge locations. They suspect that there are too many distinct descriptions to address without expert knowledge and that the Description data should be reflected in the Charge Code entries.

In their pursuit of the Hot Xs, Alex and Alice will use Pareto plots, tree maps, and the data filter, together with dynamic linking.

4.4.1. Exploring Two Unusual Accounts

At this point, it seems reasonable for Alice to construct Pareto plots for each of these variables: Account, Charge Code, and Charge Location. But, to gauge the effect of each variable on late charges, she and Alex decide to weight each variable by Abs(Amt), as this gives a measure of the magnitude of impact on late charges.

Alice closes all data tables and reports other than LateCharges.jmp. Recall that there were 389 different accounts represented in the data. To construct a Pareto plot, Alice selects Graph > Pareto Plot. She inserts Account as Y, Cause and Abs(Amt) as Weight. When she clicks OK, she sees a Pareto chart with 389 bars, some so small that they are barely visible (see Exhibit 4.47).

Figure 4.47. Pareto Plot for Account

She realizes that these barely visible bars correspond to Accounts with very ...