Concept | Variable types for interactive statistics#

Recognizing the different variable types in a worksheet can help you to identify the appropriate statistics to compute on them, as well as to understand certain behaviors of Dataiku.

Categorical and numerical variables#

In Dataiku the character \(A\) next to a variable name indicates that it’s categorical (nominal or ordinal), while the character \(\#\) indicates that it’s numerical.

Variables in statistics cards#

Also, Dataiku enforces the use of the correct variable type when computing statistics. For instance, when creating a PCA card or a correlation matrix card, Dataiku disables the selection of categorical variables so that these cards don’t return meaningless results.

Likewise, when you choose variables for cards such as the univariate and bivariate analysis cards, Dataiku automatically selects the appropriate statistics options to compute. It disables the selection of options that would be meaningless for the types of variables that you selected.

Note

After creating a univariate or bivariate card, you can change the way that Dataiku treats the variables used in the card. However, this change doesn’t affect the schema of the dataset or the default way that Dataiku will treat the same variables in a different card.