This article describes how to save new variables containing predicted probabilities of each response from a machine learning model or ensemble. The below table consists of respondent level probability scores for brand preference.
- A Machine Learning output (e.g., CART, Support Vector Machine, Linear Discrimination Analysis, etc.)
- Select the Machine Learning output.
- Go to the object inspector > Inputs > SAVE VARIABLE(S) > Probabilities of Each Response, or go to Anything > Advanced Analysis > Machine Learning > Save Variables(s) > Probabilities of Each Response. A variable set will appear in the Data Sets tree.
- OPTIONAL: To create a table showing respondent level values go to the Anything menu and select Tables > Raw Data > Variables and then in the object inspector under Inputs >RAW DATA > Variables select the variables created above.