GLMPerSegment Input - 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
previous
Product Category
Teradata Vantageā„¢
Table Description
Input table Contains data sets to use to build models. Function builds one model for each partition.
AttributeTable [Optional] Contains predictors to use for specified partitions.

If any input table partition key (which may have one or more columns) is not in AttributeTable, function uses all input table attributes specified by TargetColumns for that partition to build the model.

If input table partition key is in AttributeTable, function replaces attributes specified by TargetColumns with attributes specified in AttributeTable.

If you omit AttributeTable, function uses every specified target column (every numeric_input_column and categorical_input_column) in every model.

ParameterTable [Optional] Contains model parameter values to use for specified partitions.

If any input table partition is not in ParameterTable, function uses parameter values specified by syntax elements or default parameter value to build the model.

If input table partition key is in ParameterTable, each parameter value in ParameterTable overrides corresponding parameter value specified by either syntax element or default.

If you omit ParameterTable, function uses parameter values specified by syntax elements for every partition.

InputTable Schema

Column Data Type Description
partition_column Any allowed by Partition Key [Column appears once for each specified partition_column.] Column by which input table is partitioned. Function creates one model for each partition.
response_column Any numeric SQL data type Dependent/response variable. If column contains NULL, function ignores row.
numeric_column Any numeric SQL data type [Column appears one or more times.] Independent/predictor variable. If column contains NULL, function ignores row.
categorical_column Any [Column appears once for each categorical_column.] Categorical independent/predictor variable. If column contains NULL, function ignores row.

AttributeTable Schema

Column Data Type Description
partition_column Any allowed by Partition Key [Column appears once for each specified partition_column.] Column on which to partition table.

You must specify this column in PARTITION BY clause of input table.

Partition columns must start in first column, be contiguous, and appear before attribute_column and value_column in the table.

attribute_column VARCHAR Contains names of target columns in input table.

If this column contains duplicate values, the function returns an error.

value_column INTEGER 1 for numeric attribute, 0 for categorical attribute.

ParameterTable Schema

Column Data Type Description
partition_column Any allowed by Partition Key [Column appears once for each specified partition_column.] Column on which to partition table.

Partition columns must start in first column, be contiguous, and appear before parameter_column and value_column in the table.

parameter_column VARCHAR One of the following syntax elements:
  • Family
  • FitMethod
  • Alpha
  • RegularizationLambda (or Lambda)
  • StopThreshold
  • MaxIterNum
  • FeatureScale
  • CategoricalEncoding
  • MinSamplesForEncoding
  • Smoothing

If this column contains duplicate values, the function returns an error.

value_column VARCHAR Value to use for syntax element for partition.