Setup and Requirements | teradataml open-source machine learning functions - Setup and Requirements - Teradata Package for Python

Teradata® Package for Python User Guide

Deployment
VantageCloud
VantageCore
Edition
Enterprise
IntelliFlex
VMware
Product
Teradata Package for Python
Release Number
20.00
Published
December 2024
Language
English (United States)
Last Update
2024-12-18
dita:mapPath
nvi1706202040305.ditamap
dita:ditavalPath
plt1683835213376.ditaval
dita:id
rkb1531260709148
Product Category
Teradata Vantage
teradataml open-source machine learning functions support the following Vantage platforms:
  • VantageCloud Enterprise
  • VantageCloud Lake

Setup and requirement information common to both platforms follows.

Use of Script and Apply Table Operators

When both table operators (Script and Apply) are available, you can execute feature functions using either operator:
  • For systems with Teradata in-nodes Python packages (STO), use the Script table operator:
    from teradataml import configure
    configure.table_operator = "script"
    
  • For systems with the Apply table operator:
    from teradataml import configure
    configure.table_operator = "apply"
    

If no configuration option is specified, teradataml checks the database version. If the version is 20 or higher, it will execute using the Apply table operator; otherwise, it will default to the Script table operator.

Version requirement

teradataml open-source machine learning functions support the classes, methods, and attributes of the classes in the underlying open source machine learning package if the underlying package version in Vantage and the version on the client are the same.

For example, in scikit-learn, for the class OneVsRestClassifier, an attribute intercept_ is supported on scikit-learn v0.24.2, whereas it is not supported on v1.1.3.

So, when you run the following:
obj = OneVsRestClassifier(...)

obj.fit(X, y)

obj.intercept_
This works fine with scikit-learn v0.24.2, but fails on scikit-learn v1.1.3 with AttributeError.

Same behavior will be seen with teradataml open-source machine learning functions.