Predictor Screening Results - Binary Response Tab
The Binary response tab will be present in the Predictor Screening Results dialog box only if a binary variable was selected in the Predictor Screening Startup Panel.
| Element Name | Description |
|---|---|
| Summary | |
| Variable importance | Click this button to produce a variable importance spreadsheet with the predictors sorted according the criterion specified on the Advanced tab of the Predictor Screening Startup Panel. |
| Results | This group box contains options to display the results of the predictor screening analysis for the selected binary response variable. |
| Select predictor | From this drop-down list, select a specific predictor to display results for, or select All predictors to display output for all selected predictors. |
| Crosstabulation | Click this button to create a spreadsheet that contains the crosstabulation between the binary response and the selected optimally binned predictor variable.
Along with the frequencies, Statistica displays important statistical information for each optimal bin, including weight of evidence and information value, which enable you to understand how the distribution of good and bad values change across levels of the optimally binned predictor variable. |
| Continuous predictors | This group box contains options to create two graphs specifically for continuous predictors. |
| Distribution | Click this button to create an overlaid graph of the empirical cumulative distribution function, smoothed probability density function, and the default rates.
The empirical cumulative distribution function and the smoothed probability density function are based on the original predictor variable. The default rates are associated with the optimal bins discovered by Statistica . The default rates appear as red horizontal lines. The width of each line spans the interval for the optimally binned predictor and the height of the line is the default rate for the associated bin. |
| Default rates | Click this button to produce a scatterplot of the default rates and the original predictor variable.
Statistica treats each unique value of the original predictor variable as its own bin, and the default rate is then computed for the unique value. The unique value of the predictor variable is plotted on the x-axis, and its associated default rate is plotted along the y-axis. Additionally, the default rates for the optimally binned predictor variable are overlaid on top of the scatterplot. The default rates appear as red horizontal lines. The width of each line spans the interval for the optimally binned predictor and the height of the line is the default rate for the associated bin. |