Sensitivity of models in the Hall of Fame
Model sensitivity is the correlation between behavior predicted by the predictive model, and the behavior predicted by one of its predictors.
- In the Hall of Fame step, click the model name link.
- In the Model inspection dialog box, click the Sensitivity tab.
The correlation of predictors with the predicted behavior is displayed as a bar chart. Consider the level of correlation of the predictors on the right side:
- If they are very low (below 0.01), decrease the size of the models. In theGrouping predictors step, adjust the number of groups and remove the groups that contain the low correlation predictors.
- If they are high (above 0.1), increase the size of the models. In theGrouping predictors step, adjust the number of groups and increase the grouping level.
Related information
Open topic with navigation