1.1 - 8.10 - Identification of Numeric and Categorical Columns - Teradata Vantage

Teradata Vantage™ - Machine Learning Engine Analytic Function Reference

Teradata Vantage
Release Number
October 2019
Content Type
Programming Reference
Publication ID
English (United States)
To work properly, these functions require you to identify the columns to treat as categorical input variables:
  • CoxPH
  • DecisionForest
  • DecisionForestPredict_MLE
  • GLM
  • GLML1L2
  • Naive Bayes Classifier
  • NaiveBayesPredict_MLE
  • XGBoost

If a column contains nonnumeric values, or numeric values that have no numeric relationship, identify the column as categorical.

For example, in the following table, the numeric values in the columns Area and Population have a numeric relationship—a higher number indicates a higher value. Therefore, Area and Population are numeric inputs. The numeric values in the column Region have no numeric relationship; they are labels. You must identify Region as a categorical column.

State Area (Thousand Sq. Mi.) Population (Millions) Region
New York 54.60 19.7 1
Pennsylvania 46.00 12.8 1
Georgia 59.40 10.3 2
Alabama 52.40 4.9 2
Iowa 56.30 3.1 3
Illinois 57.90 12.8 3

You must identify a column of binary values as categorical, whether its values are nonnumeric (such as 'T' and 'F') or numeric (1 and 0).