Changes and Additions - Teradata Vantage

Machine Learning Engine Analytic Function Reference
Product
Teradata Vantage
Release Number
8.10
1.1
Published
October 2019
Language
English (United States)
Last Update
2019-12-31
dita:mapPath
ima1540829771750.ditamap
dita:ditavalPath
jsj1481748799576.ditaval
dita:id
B700-4003
lifecycle
previous
Product Category
Teradata Vantage™
Date Release Description
October 2019 MLE 8.10.01 Minor revisions and corrections.
July 2019 MLE 8.10 Names of some functions, syntax elements, input tables, and output tables were standardized. See Function Name Changes for Vantage 1.1 and Table Aliases and Syntax Elements in Error Messages.
March 2019 MLE 8.10 Every classifier function now has the following:
  • Accumulate syntax element for specifying input columns to copy to the output table
  • Ability to output probability that observation belongs to predicted class
This enhancement affected these functions:
March 2019 MLE 8.10 Each of these predictive modeling functions now has a CategoricalEncoding syntax element, which improves its ability to handle categorical columns with many unique levels:
March 2019 MLE 8.10 CCMPrepare (ML Engine) function output table column name aster_ccm_id is now ccm_id.
March 2019 MLE 8.10 GLM2 and GLM2Predict functions have been deprecated with this release and replaced with GLML1L2 (ML Engine) and GLML1L2Predict (ML Engine). GLML1L2 supports the Gaussian and binomial (logistic) distributions. Unlike GLM2, it does not support the Poisson distribution.
March 2019 MLE 8.10 KMeans (ML Engine) function output table has a changed column name for UnpackColumns ('false'). See KMeans Output.
March 2019 MLE 8.10 KNN (ML Engine) function has these enhancements:
  • PartitionColumn syntax element lets you to specify a column with a unique identifier for each row, resulting in more uniform partitioning and better performance.
  • Omitting PartitionBlockSize syntax element causes function to try to calculate optimal value.
March 2019 MLE 8.10 "ASTER" was removed from some token names in model table created by NaiveBayesTextClassifierTrainer (ML Engine) function.
March 2019 MLE 8.10 RandomSample (ML Engine) function now has SetIDAsFirstColumn syntax element syntax element that lets you specify where set_id column appears in output table.
March 2019 MLE 8.10 Receiver Operating Characteristic (ROC) (ML Engine) function output now depends on whether you specify the syntax element OutputTable or ROCTable. See ROC Output.
September 2018 MLE 8.00.01 Initial release.