7.00.02 - Output - Aster Analytics

Teradata Aster® Analytics Foundation User GuideUpdate 2

Product
Aster Analytics
Release Number
7.00.02
Published
September 2017
Content Type
Programming Reference
User Guide
Publication ID
B700-1022-700K
Language
English (United States)
Last Update
2018-04-17

Onscreen Output

Column Description
dfDevRatio Number of nonzero coefficients for DevRatio in each lambda value.
DevRatio Fraction of (null) deviance explained. Deviance calculations incorporate weights if present in the model. Therefore, this value is (1-deviance)/(null deviance). (For elastic net, this value is the R-square.)
dfDev Number of nonzero coefficients for deviance in each lambda value.
deviance Deviance, which is defined as 2*(loglike_sat - loglike), where loglike_sat is the log-likelihood for the saturated model (a model with a free parameter for each observation).
lambda Lambda value.

Model Table

GLM2 Model Table Schema
Column Data Type Description
group VARCHAR Contains the abstract categories for output, which are:
  • A: global properties
  • B: regularization-specific properties
  • C: coefficients
information VARCHAR Contains the values provided in either this column or the value column (for example, FAMILY:POISSON).
value VARCHAR, INTEGER, or DOUBLE PRECISION Contains the value of each item specified in the information column.

Regularization Table

The function outputs this table only if you specify the RegularizationTable argument.

GLM2 Regularization Table Schema
Column Data Type Description
df_dev_ratio DOUBLE PRECISION Number of nonzero coefficients for %dev in each lambda value.
deviance_ratio DOUBLE PRECISION Fraction of (null) deviance explained. Deviance calculations incorporate weights if present in the model. Therefore, this value is (1-deviance)/(null deviance). (For elastic net, this value is the R-square.)
df_dev DOUBLE PRECISION Number of nonzero coefficients for deviance in each lambda value.
deviance DOUBLE PRECISION Deviance, which is defined as 2*(loglike_sat - loglike), where loglike_sat is the log-likelihood for the saturated model (a model with a free parameter for each observation).
lambda DOUBLE PRECISION Lambda value.
predictor_variable_column DOUBLE PRECISION Coefficients for all predictors for the current lambda value.