This document provides detailed description and complete usage information for all the functions in the Teradata® Package for Python, teradataml.
teradataml Function Categories
Categories of teradataml package functions:
- <name>: Analytic functions available to use as class. Executing each analytical function means, generating instance of the analytic function class.
- DataFrame.<name>: Create a teradataml DataFrame for the underlying Teradata table.
- <name>_context: API functions that manage the connection and certain internal data structures called context.
- <TDMLDF>.<name>: teradataml DataFrame methods available for data manipulation, preparation and exploration.
Notes:
- name is part of a function name that indicates the specific task of the function.
- TDMLDF is teradataml DataFrame object.
teradataml Analytic Function Default Execution Locations
The teradataml analytics package includes two subpackages:
- teradataml.analytics.mle
Analytic functions in the teradataml.analytics.mle subpackage require your system to have the Vantage Machine Learning Engine, which is a separate machine learning legacy engine that is not part of the current standard Vantage offer. If your Vantage system does not have the required ML Engine, an error or no-op behavior will occur when functions in this subpackage are invoked. - teradataml.analytics.sqle
All teradataml analytic functions are in either of these two subpackages.
For Vantage 1.0 the following nine teradataml analytic functions were executed by default in the Advanced SQL Engine:
- Attribution
- DecisionForestPredict
- DecisionTreePredict
- GLMPredict
- NaiveBayesPredict
- NaiveBayesTextClassifierPredict
- NPath
- Sessionize
- SVMSparsePredict
For Vantage 1.1 or later versions, six new functions are added to the list. The following 15 teradataml analytic functions are executed by default in the Advanced SQL Engine:
- Antiselect
- Attribution
- DecisionForestPredict
- DecisionTreePredict
- GLMPredict
- MovingAverage
- NaiveBayesPredict
- NaiveBayesTextClassifierPredict
- NGramSplitter
- NPath
- Pack
- Sessionize
- StringSimilarity
- SVMSparsePredict
- Unpack
All other teradataml analytic functions are executed by default in the Teradata Machine Learning Engine.
Notes:
-
Teradata recommends importing an analytic function in one of the following ways:
- Preferred: Import from the teradataml package.
- To choose the engine where the analytic function is used: Import from the subpackage.
-
The teradataml package includes a module (load_example_data) with datasets for the examples in analytic functions. To execute these examples, you need the following:
- To have a connection to Teradata Vantage
- Import the load_example_data module to load the data, using the command:
from teradataml.data.load_example_data import load_example_data