TD_GLMPredict Examples | GLMPredict | Teradata Vantage - Examples: How to Use TD_GLMPredict - Teradata Vantage

Example: TD_GLMPredict Using Credit Data

This example takes credit data and uses TD_GLM function to get a model. You can view the input and output in the TD_GLM example.

TD_GLMPredict Call for Credit Data

CREATE VOLATILE TABLE vt_glm_predict_credit_ex AS (
    SELECT * from TD_GLMPredict (
      ON credit_ex_merged AS INPUTTABLE
      ON td_glm_output_credit_ex AS Model DIMENSION
      USING
      IDColumn ('ID')
      Accumulate('Outcome')
    ) AS dt
) WITH DATA
ON COMMIT PRESERVE ROWS;

TD_GLMPredict Output for Credit Data

Example: TD_GLMPredict Using Housing Data

You can view the input and output of steps 1 through 3 in the TD_GLM example.

TD_GLMPredict Call for Housing Data

CREATE VOLATILE TABLE vt_predict_cal_ex AS (
    SELECT * from TD_GLMPredict (
      ON cal_housing_ex_scaled AS INPUTTABLE
      ON td_glm_cal_ex AS Model DIMENSION
      USING
      IDColumn ('ID')
      Accumulate('MedHouseVal')
    ) AS dt
) WITH DATA
ON COMMIT PRESERVE ROWS;

TD_GLMPredict Output for Housing Data

TD_GLMPredict Call with Family: Gaussian

SELECT * FROM TD_GLMPredict (
  ON housing_train_segment AS InputTable PARTITION BY partition_id
  ON (SELECT * FROM TD_GLM (
     ON housing_train_segment AS InputTable PARTITION BY partition_id ORDER BY sn
     USING
     Family('Gaussian')
     InputColumns('[3:10]')
     ResponseColumn('price')
     BatchSize(10)
     MaxIterNum (1000)
     ) AS output_glmsegment_gaussian) AS ModelTable PARTITION BY partition_id
  USING
  IDColumn ('sn')
  Accumulate('price')
) AS dt ORDER BY 1, 2;

TD_GLMPredict Output with Family: Gaussian

TD_GLMPredict Call with Family: Binomial

SELECT * from TD_GLMPredict (
  ON housing_train_segment AS InputTable PARTITION BY partition_id
  ON (select * from TD_GLM (
    ON housing_train_segment AS InputTable partition by partition_id
    USING
    Family('Binomial')
    InputColumns('[3:10]')
    ResponseColumn('homestyle')
    MaxIterNum (100)
    ) as glm_output) AS ModelTable PARTITION BY partition_id
  USING
  IDColumn ('sn')
  OutputProb('true')
  Responses('0','1')
  Accumulate('homestyle')
) AS dt order by 1, 2;

TD_GLMPredict Output with Family: Binomial