Correlation Syntax Elements - Teradata Vantage

Machine Learning Engine Analytic Function Reference

Product

Teradata Vantage

Release Number

9.02

9.01

2.0

1.3

Published

February 2022

Language

English (United States)

Last Update

2022-02-10

dita:mapPath

rnn1580259159235.ditamap

dita:ditavalPath

ybt1582220416951.ditaval

dita:id

B700-4003

lifecycle

Product Category

Teradata Vantage™

VIF

[Optional] Specify whether the function computes the variance inflation factor (VIF) score of each attribute specified by TargetColumns. The VIF score for an attribute indicates how strongly the attribute correlates with other attributes. The higher the VIF score, the stronger the correlation.

The function computes each VIF score iteratively. For each iteration, the function removes the attribute with the strongest collinearity from the data and computes the VIF score for the remaining attributes. Iterations continue until all multicollinear attributes are removed. VIFThreshold specifies the criteria for multicollinearity.

Default: 'false'

TargetColumns

[Required with VIF ('true'), disallowed otherwise.] Specify the names of the target columns for which to compare the VIF with the specified vif_threshold.

TargetColumnPairs

[Required with VIF ('false'), disallowed otherwise.] Specify pairs of columns for which to calculate correlations. For each column pair, 'col_name1:col_name2', the function calculates the correlation between col_name1 and col_name2. For each column range, '[col_index1:col_index2]', the function calculates the correlation between every pair of columns in the range. For example, if you specify '[1:3]', the function calculates the correlation between the pairs (1,2), (1,3), (2,3),(1,1),(3,3). The minimum value of col_index1 is 0, and col_index1 must be less than col_index2.

PartitionColumns

[Optional] Specify the names of the input columns that define the partition columns. The function computes correlation and VIF scores for each partition separately.

Default behavior: All input columns belong to a single partition.

VIFThreshold

[Optional with VIF ('true'), disallowed otherwise.] Specify the VIF threshold, a DOUBLE PRECISION value of at least 1. To detect significant collinearity, specify a vif_threshold in [5, 20].

Default: 10

ExceptionAttribute

[Optional with VIF ('true'), disallowed otherwise.] Specify the name of an InputTable column, also specified by TargetColumns, that the function cannot label multicollinear during VIF calculation. If exception_attribute is not also a target_column, an error occurs.

OutputSummary

[Optional with VIF ('true'), disallowed otherwise.] Specify whether to summarize the multicollinear output:

Option	Description
'true'	Show only final result of algorithm that calculates VIF score (one label for each partition). Use this value to show final results (which attributes are identified as multicollinear).
'false'	Show results for each iteration of algorithm that calculates VIF score. Use this value to show VIF score at end of each iteration, which can help you interpret final results.

Default: 'true'