NaiveBayesTextClassifierPredict Example - Teradata Vantage

Input

• Input table: predicts, created by applying the TextTokenizer function to complaints_test, a log of vehicle complaints
• Model Table: complaints_tokens_model, output by NaiveBayesTextClassifierTrainer Example

SQL Call

SELECT * FROM NaiveBayesTextClassifierPredict@coprocessor (
  ON (
    SELECT doc_id, lower(token) AS token
    FROM TextTokenizer (
      ON complaints_test PARTITION BY ANY
      USING
      TextColumn ('text_data')
      OutputByWord ('true')
      Accumulate ('doc_id')
    ) AS dt1
  ) AS predicts PARTITION BY doc_id
  ON complaints_tokens_model AS model DIMENSION
  USING
  InputTokenColumn ('token')
  ModelType ('Bernoulli')
  DocIDColumns ('doc_id')
  TopK (1)
) AS dt;

Output