1.0 - 8.00 - XGBoostPredict Example 3: Sparse Input Format - Teradata Vantage

Teradata® Vantage Machine Learning Engine Analytic Function Reference

Product
Teradata Vantage
Release Number
1.0
8.00
Release Date
May 2019
Content Type
Programming Reference
Publication ID
B700-4003-098K
Language
English (United States)

Input

  • input: sparse_housing_test_binary, created by inputting housing_test_binary (input table for XGBoostPredict Example 1: Binary Classification) to the Unpivoting function:
    DROP TABLE IF EXISTS sparse_housing_test_binary;
    
    CREATE MULTISET TABLE sparse_housing_test_binary AS (
      SELECT sn, homestyle, attribute, value_col AS value1 FROM 
        Unpivoting (
          ON housing_test_binary
          USING
          TargetColumns (
            'price','lotsize','bedrooms','bathrms','stories',
            'garagepl','driveway','recroom','fullbase','gashw','airco','prefarea'
          )
          Accumulate ('sn','homestyle')
        ) AS dt
    ) WITH DATA;
    
  • ModelTable: sparse_housing_xgboost_model, output by XGBoost Example 3: Sparse Input Format
input: sparse_housing_test_binary
sn homestyle attribute value1
13 Classic price 27000.0
13 Classic lotsize 1700.0
13 Classic bedrooms 3
13 Classic bathrms 1
13 Classic stories 2
13 Classic garagepl 0
13 Classic driveway yes
13 Classic recroom no
13 Classic fullbase no
13 Classic gashw no
13 Classic airco no
13 Classic prefarea no
25 Classic price 42000.0
25 Classic lotsize 4960.0
25 Classic bedrooms 2
25 Classic bathrms 1
25 Classic stories 1
25 Classic garagepl 0
25 Classic driveway yes
25 Classic recroom no
25 Classic fullbase no

SQL Call

CREATE MULTISET TABLE sparse_housing_predict AS (
  SELECT * FROM XGBoost_Predict (
    ON sparse_housing_test_binary AS "input" PARTITION BY sn
    ON sparse_housing_xgboost_model AS modeltable DIMENSION ORDER BY tree_id, iter, class_num
    USING
    AttributeNameColumn ('attribute')
    AttributeValueColumn ('value1')
    IdColumn ('sn')
    NumBoostedTrees (1)
    Accumulate ('homestyle')
  ) AS dt
) WITH DATA;

Output

sparse_housing_predict
sn homestyle prediction confidence_lower confidence_upper
13 Classic Classic 1 1
16 Classic Classic 1 1
25 Classic Classic 1 1
38 Electic Electic 1 1
53 Electic Electic 1 1
111 Classic Classic 1 1
117 Electic Electic 1 1
132 Classic Classic 1 1
140 Classic Classic 1 1
142 Classic Classic 1 1
157 Electic Electic 1 1
161 Electic Electic 1 1
176 Electic Electic 1 1
177 Electic Electic 1 1
195 Classic Classic 1 1
198 Classic Classic 1 1
224 Electic Electic 1 1
234 Classic Classic 1 1
237 Classic Classic 1 1
239 Classic Classic 1 1
249 Classic Classic 1 1
251 Classic Classic 1 1
254 Electic Electic 1 1
255 Electic Electic 1 1
260 Classic Classic 1 1
274 Electic Electic 1 1
294 Classic Classic 1 1
301 Electic Electic 1 1
306 Electic Electic 1 1
317 Electic Electic 1 1
340 Electic Electic 1 1
353 Electic Electic 1 1
355 Electic Electic 1 1
364 Electic Electic 1 1
401 Electic Electic 1 1
403 Electic Electic 1 1
408 Electic Electic 1 1
411 Electic Electic 1 1
440 Electic Electic 1 1
441 Electic Electic 1 1
443 Electic Electic 1 1
459 Classic Classic 1 1
463 Classic Classic 1 1
469 Electic Electic 1 1
472 Electic Electic 1 1
540 Electic Electic 1 1