2.0 - PMMLPredict Input - Teradata Vantage

Teradata Vantageā„¢ - Bring Your Own Model User Guide

Product
Teradata Vantage
Release Number
2.0
Release Date
October 2021
Content Type
User Guide
Publication ID
B700-1111-051K
Language
English (United States)
Input Table Description
InputTable Columns contain data to score.

InputTable typically has same schema as data used to train model.

It must have all columns that model requires (and may have additional columns). Columns that model requires must have same names in InputTable (case-insensitive) and compatible data types (see following table), but can be in different order.

ModelTable Represents one or more models to use for scoring. Has one row for each model.

One PMMLPredict function call can score only one model. If ModelTable has multiple models, specify model_id of model to score in WHERE clause.

Neither InputTable nor ModelTable can have a column named 'dedicated' or 'json_result'.
Compatible ModelTable and InputTable Data Types
ModelTable Column Data Type InputTable Column Data Type
Numeric BYTEINT

SMALLINT

INTEGER

BIGINT

DECIMAL or NUMERIC

FLOAT, REAL, or DOUBLE PRECISION

NUMBER

Categorical CHAR

VARCHAR

ModelTable Schema

Column Data Type Description
model_id VARCHAR Model identifier.

This column must appear in PRIMARY INDEX clause of CREATE SET TABLE statement.

model BLOB PMML model.

ModelTable has one row for each model.