In-Database Analytic Functions - Teradata Warehouse Miner

In-Database Analytic Functions User Guide

Product
Teradata Warehouse Miner
Release Number
5.4.5
Published
February 2018
Language
English (United States)
Last Update
2018-05-04
dita:mapPath
dfw1503087325991.ditamap
dita:ditavalPath
ft:empty
dita:id
B035-2306
lifecycle
previous
Product Category
Teradata® Warehouse Miner

This document describes a number of in-database analytic functions that are invoked from any Teradata SQL query execution product, such as Teradata Studio or Teradata Bteq. The controlling program for these analytic functions is a custom Teradata External Stored Procedure called TD_Analyze. Some of the functions also use custom Teradata Table Operators. To use the Teradata Warehouse Miner (Miner) In-Database Analytic Functions, these user-defined functions (UDF) must be installed in the Teradata database using a console application supplied with Miner, accessible from the Start menu in the Miner program group. Once these UDFs are installed, they are executed without invoking the Miner client application.

Although independent of the Miner client application, the in-database analytic functions are based on functions available in Miner. All of the in-database functions and more are provided in the Miner client application, which has three principal function categories.

Function Category Description
Data Profiling Descriptive statistics provided to generate reports and graphics with drill down capabilities, pointing out potential data quality issues. The In-Database Descriptive Statistics functions are derived from the corresponding functions in the Data Profiling function category.
Analytic Data Set (ADS) Generation Functions used to build and transform analytic data sets. The In-Database Variable Transformation function is derived from the Variable Transformation function in the Analytic Data Set Generation function category.
Analytic Functions Functions used to invoke analytic algorithms and scoring along with statistical tests. The In-Database Fast K-Means algorithm and the Gain Ratio Extreme algorithm provide new variations of the Clustering and Decision Tree algorithms and scoring in Miner in the Analytic Functions category.
Each of the in-database functions described in this document has a counterpart that resides in the Miner client application. Graphing and reporting capabilities are included in the Minor client application, however, other features may not be available.
By default, when an in-database analytic function creates a table, the existing table is automatically dropped first to allow re-execution of the analysis. The automatic dropping of output tables prior to execution may be inhibited if the overwrite parameter is set to false (overwrite=false).
Examples in this document can be copied and used on a system where the UDF is installed and the twm default user name along with the rest of the tutorial environment (including the twm_source, the twm_result database, and the tutorial tables installed in twm_source) is set up.