Cumulative Gains and Lift Charts

Figure 1. Cumulative gains chart
Cumulative gains chart

The cumulative gains chart shows the percentage of the overall number of cases in a given category "gained" by targeting a percentage of the total number of cases. For example, the first point on the curve for the Total service category is approximately at (10%, 20%), meaning that if you score a dataset with the network and sort all of the cases by predicted pseudo-probability of Total service, you would expect the top 10% to contain approximately 20% of all of the cases that actually take the category Total service. Likewise, the top 20% would contain approximately 30% of the defaulters, the top 30% of cases, 50% of defaulters, and so on. If you select 100% of the scored dataset, you obtain all of the defaulters in the dataset.

The diagonal line is the "baseline" curve; if you select 10% of the cases from the scored dataset at random, you would expect to "gain" approximately 10% of all of the cases that actually take any given category. The farther above the baseline a curve lies, the greater the gain.

Figure 2. Lift chart
Lift chart

The lift chart is derived from the cumulative gains chart; the values on the y axis correspond to the ratio of the cumulative gain for each curve to the baseline. Thus, the lift at 10% for the category Total service is approximately 20%/10% = 2.0. It provides another way of looking at the information in the cumulative gains chart.

Note: The cumulative gains and lift charts are based on the combined training and testing samples.