1.0 - 8.00 - ShapeletUnsupervised Input - 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)

InputTable Schema

Column Data Type Description
id Any Time series identifier.
time_column BYTEINT, SMALLINT, INTEGER, BIGINT, NUMERIC, or DOUBLE PRECISION Time axis of data. Used for saxification. Corresponds to SAX input table time_column.
target_column BYTEINT, SMALLINT, INTEGER, BIGINT, NUMERIC, or DOUBLE PRECISION Time series data to cluster. Used for saxification. Corresponds to SAX input table target_column.

Correct Input Time Series Data Format

If input time series data is incorrectly formatted, function behavior is undefined. In a correctly formatted time series, time intervals are evenly spaced and all time intervals have numeric values.

To calculate missing values in a time series, use the Interpolator function.

If the input table time column is text-based, create a new input table with an integer-based time column. For example, suppose that the table time_series_text_time has the text-based columns idval, timeval, valueval, and catval. This statement creates a table that the function accepts as input:

CREATE MULTISET TABLE time_series_numeric_time AS (
  SELECT
    idval,
    RANK() OVER (PARTITION BY idval ORDER BY timeval) AS timeval,
    valueval,
    catval
    FROM time_series_text_time
  ) WITH DATA;