This article describes how to create a simulator for your predictive model. This is sometimes referred to as a "typing tool". The simulator allows you to enter new values for the Predictor variables in your model and to see what the predicted value is for that combination, based on your model.
Requirements
A regression or machine learning model output created via Anything > Advanced Analysis > Regression or Machine Learning.
In this example, we will use a Gradient Boosting using an Outcome variable of "Preferred cola" and Predictor variables of "Age", "Income", "Gender" and "Living Arrangements".
Method
1. Select your predictive model output.
2. Click Diagnostics > Create Simulator in the object inspector.
3. A new page will be added that includes:
- Combo boxes for each of the predictor variables.
- A probability table of the predicted outcome category.
- The final predicted outcome category.
4. OPTIONAL: Format and arrange the labels, controls, and outputs from the simulator to your preferred appearance via the Appearance tab for the calculations and the Control tab for the combo boxes.
Note the following:
- When the Outcome variable in your model is numeric, the simulator will provide one output showing the predicted value for your selection. When your Outcome variable is categorical, the simulator will create two outputs, with the first showing the predicted category, and the second showing the probabilities of the possible outcomes for your selection.
- If you have categorical Predictor variables in your model that have empty categories you will not be able to simulate values for those categories. You can remove these ahead of time (before creating the simulator) by creating a table of the relevant variable set, selecting the label for the empty category, and pressing Delete, or you can simply remove those labels from the simulator once it has been added to the page, by selecting the combo box, going in to Control > Item list on the right and deleting the label for the empty category.
Next
How to Do Gradient Boosting Analysis
How to Run a Gradient Boosting Machine Learning Model
How to Run Support Vector Machine