Scoring Wizard
You can use the Scoring Wizard to apply a model created with one dataset to another dataset and generate scores, such as the predicted value and/or predicted probability of an outcome of interest.
Example. The direct marketing division of a company uses results from a test mailing to assign propensity scores to the rest of their contact database, using various demographic characteristics to identify contacts most likely to respond and make a purchase.
To score a dataset with a predictive model
- Open the dataset that you want to score.
- Open the Scoring Wizard. From the menus choose:
- Select a model XML file or compressed file archive (.zip fle). Use the Browse button to navigate to a different location to select a model file.
- Match fields in the active dataset to fields used in the model. See the topic Scoring Wizard: Matching model fields to dataset fields for more information.
- Select the scoring functions you want to use. See the topic Scoring Wizard: Selecting scoring functions for more information.
Select a Scoring Model. The model file can be an XML file or a compressed file archive (.zip file) that contains model PMML. The list only displays files with a .zip or .xml extension; the file extensions are not displayed in the list. You can use any model file created by IBM® SPSS® Statistics. You can also use some model files created by other applications, such as IBM SPSS Modeler, but some model files created by other applications cannot be read by IBM SPSS Statistics, including any models that have multiple target fields (variables).
Model Details. This area displays basic information about the selected model, such as model type, target (if any), and predictors used to build the model. Since the model file has to be read to obtain this information, there may be a delay before this information is displayed for the selected model. If the XML file or .zip file is not recognized as a model that IBM SPSS Statistics can read, a message indicating that the file cannot be read is displayed.