This command is available with the Predictive Analytics Module. Click here for more information about how to activate the module.

Use the partial dependence plots to gain insight into how the
important variables or pairs of variables affect the estimated event
probabilities of the predicted response.

Minitab plots the marginal mean of the 1/2 log odds versus each predictor to
help identify the range of predictor values that produce higher event
probabilities. This one-to-one relationship between fit = ½ log (event prob/1 –
event prob) and event probability is easy to interpret. For example, from the
chart, when fit = 0, the event probability = 0.5.
###### Note

To add more partial dependence plots, click
Select More Predictors to Plotafter
the last plot of that type in the results.

Odds = event prob / (1-event prob)

The one predictor partial dependence plot displays how the average fit which represents the event probability changes with changes in the predictor levels.

The two predictor partial dependence plot shows the interaction effects of the plotted predictors on the fits. Because of the relationship between the fits and the event probability, you can use this plot to help identify optimal predictor values. Event probability monotonically increases as fits increases.

The two predictor partial dependence plot indicates how the response will change with changes in the predictor levels of two important variables. For categorical predictors, Minitab displays a matrix plot of the various relationships at the various levels of the predictors. For continuous predictors, Minitab displays a surface plot or a contour plot of this relationship.