XGBoostPredict Example 2: Multiple-Class Classification - Teradata Vantage

Input

• input: iris_test, created in DecisionTree Example 1
• ModelTable: xgboost_model2, output by XGBoost Example 2: Multiple-Class Classification

SQL Call

CREATE MULTISET TABLE iris_predict AS (
  SELECT * FROM XGBoostPredict (
    ON iris_test AS "input" PARTITION BY ANY
    ON xgboost_model2 AS modeltable DIMENSION ORDER BY tree_id, iter, class_num
    USING
    IDColumn ('id')
    NumBoostedTrees (2)
    Accumulate ('species')
  ) AS dt
) WITH DATA;

Output

The table iris_predict is similar to the following table, but has 120 rows.