1.0 - 8.00 - ShapeletSupervisedClassifier Example - 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

  • time_series: shapelets_test, which has additional data from the data set used to train the model
  • shapelets: shapelets_model, output table of ShapeletSupervised Example
time_series: shapelets_test
id period stockprice stock_category
5 22418 460 Technology
5 22419 457 Technology
5 22420 452 Technology
5 22421 459 Technology
5 22422 462 Technology
5 22423 459 Technology
5 22424 463 Technology
5 22425 479 Technology
5 22426 493 Technology
5 22427 490 Technology
... ... ... ...

SQL Call

CREATE MULTISET TABLE shapelets_predict AS (
  SELECT * FROM ShapeletSupervisedClassifier (
    ON shapelets_test AS time_series PARTITION BY id ORDER BY period
    ON shapelets_model AS shapelets DIMENSION ORDER BY shapelet_id, time_instant
  USING
  TimeInterval (1)
  TargetColumn ('stockprice')
  Accumulate ('stock_category')
  ) AS dt
) WITH DATA;

Output

This query returns the following table:

SELECT * FROM shapelets_predict ORDER BY 1;

The column stock_category contains the original category.

id predicted_category stock_category
5 Technology Technology
5 Technology Technology
5 Technology Technology
5 Technology Technology
5 Technology Technology
5 Technology Technology
5 Technology Technology
5 Technology Technology
5 Technology Technology
5 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
6 Technology Technology
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
7 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare
8 Healthcare Healthcare

Prediction Accuracy

This query returns the prediction accuracy:

SELECT (SELECT COUNT(id) FROM shapelets_predict
  WHERE predicted_category = stock_category)/
    (SELECT COUNT(id) FROM shapelets_predict) AS prediction_accuracy;
prediction_accuracy
1.00000000000000000000

The prediction accuracy is 100% because the predicted and original categories are the same.