Not all ordinal variables can be treated as continuous. If the variable you’re considering is the result of a survey where people listed their satisfaction on a scale of 1 to 5, it doesn’t make sense to try to change this variable to a continuous variable, it is an ordinal variable, and should always be treated like one. However, Altair Knowledge Studio software infers whether a numeric variable is ordinal or continuous, and sometimes a user wants to override this inference and make the other choice.
This can be done via the attribute editor. For any model in Knowledge Studio, you will be able to access the attribute editor via a button on one of the pages in the wizard. For example, in the Logistic Regression wizard, you can find the button to access the attribute editor on the Variable Selection page, as seen here:
Once you enter the attribute editor itself, look for the usage column. You can click the cells in this column to switch between eligible usage types. For example, as age is a numeric variable, it could be treated as either continuous or ordinal.
This will allow users to choose the usage type most appropriate for their model.
- Data Fabric: Stitch Up Your Data Strategy With Visualization - September 13, 2019
- Data Fabric: Save a Stitch in Time with Predictive Analytics - September 5, 2019
- Data Fabric: Skip the Patchwork with Powerful Data Prep - August 14, 2019