16.20 - NaiveBayesTextClassifierPredict Input - Teradata Database - Teradata Vantage NewSQL Engine

Teradata Vantage™ - NewSQL Engine Analytic Functions

Product
Teradata Database
Teradata Vantage NewSQL Engine
Release Number
16.20
Release Date
July 2019
Content Type
Programming Reference
Publication ID
B035-1206-162K
Language
English (United States)
Table Description
PredictorValues Contains test data, for which to predict outcomes, in document-token pairs. To transform input document into this form, input it to ML Engine function TextTokenizer or TextParser.
TextTokenizer and TextParser have language-processing limitations that might limit support for Unicode input data (see Teradata Vantage™ Machine Learning Engine Analytic Function Reference, B700-4003).
Model Model output by ML Engine NaiveBayesTextClassifierTrainer function. For schema, see Teradata Vantage™ Machine Learning Engine Analytic Function Reference, B700-4003.

PredictorValues Schema

Column Data Type Description
doc_id_column CHARACTER, VARCHAR, INTEGER, or SMALLINT [Column appears once for each specified doc_id_column.] Identifier of document that contains classified training tokens.
token_column CHARACTER or VARCHAR Classified training token.

Model Schema

For CHARACTER and VARCHAR columns, CHARACTER SET must be either UNICODE or LATIN.

Column Data Type Description
token CHARACTER or VARCHAR Classified training token.
category CHARACTER or VARCHAR Prediction category for token.
prob DOUBLE PRECISION Probability that token is in category.