Amadeus cookies policy - you'll see this message only once.

Amadeus use cookies on this website. They help us to know a little bit about you and how you use our website, which improves the browsing experience and marketing - both for you and for others. They are stored locally on your computer or mobile device. To accept cookies, continue browsing as normal. Or, go to the privacy policy for more information.

Model Comparison with SAS® Visual Statistics

SAS Visual Statistics enables you to quickly build predictive models using the following techniques:

  • Linear regression
  • Logistic regression
  • Generalised linear models
  • Clustering
  • Decision trees (available in the SAS Data Explorer)

You may decide to build multiple models to predict the same outcome as you are not sure which technique will provide the best fit. In addition to building the models, SAS Visual Statistics also provides a simple method for comparing models to select a champion model.

In this example, a loans company has built a logistic regression model and a decision tree model to predict whether a customer will default on a loan. In each case, Credit Rating, Income, Marital Status and Number of Credit Products have been included as explanatory variables in the models:

1 Logistic regression and decision tree

To choose a champion model, SAS Visual Statistics Model Comparison tool (2 model comparison tool) is used. First the models to compare are selected:

3 select models

The models to be compared must share the same response variable, level of interest (0 in this case is the value corresponding to default) and group by variable.

A new visualisation will be created showing the results of the comparison:

4 assessment plots

The plots displayed and fit statistics available depend on whether the models being compared are classification (as in this case) or continuous outcome models.  The Fit Statistic plot indicates the champion model by filling the bar blue and annotating the work “Selected” at the top. Different statistics can be displayed in each of the plots using the Options (5 options button) button:

6 change fit statistic

In this example, the logistic regression model is selected as the champion model and put into production for predicting how likely a new customer is to default on a loan.