Use the Gain and Lift charts to assess the performance of your
classification model. The Gain chart plots the total positive rate in percent
versus the percent of total counts. The Lift chart plots the cumulative lift
(or non-cumulative lift) versus the percent of total counts.
Interpretation of Gain chart
The training and test lines represent the expected response using the predictive model. The
training data set fits the model and the test data set evaluates the model. The dotted
reference line represents a line with slope = 1, which is the expected random response
without the model.
Interpretation of Lift chart
The training and test lines represent the expected response using the predictive model. The
training data set fits the model and the test data set evaluates the model. Lift is the
ratio of the gain percentage relative to the expected random result. The dotted reference
line represents a cumulative lift of 1, which means that there is no gain compared with
random.