7.00.02 - Output Table - Aster Analytics

Teradata Aster® Analytics Foundation User GuideUpdate 2

Product
Aster Analytics
Release Number
7.00.02
Release Date
September 2017
Content Type
Programming Reference
User Guide
Publication ID
B700-1022-700K
Language
English (United States)

The output table specified by the OutputTable argument stores the estimated coefficients and statistics, which are used by the functions GLMPredict and LRTEST.

When a column is not used for its corresponding row, the column contains a value of zero (0). This is a description of the columns that appear in the output table:

GLM Output Table Columns
Column Description
attribute The index of each predictor, starting from 0.
predictor The column name for each predictor that was supplied as input to the function.
category The category names of each predictor. Numeric predictors have NULL values in this column.
estimate The mean of the supplied values for each predictor.
std_error Standard deviation of the mean for each predictor (standard error).
z_score or t_score If the Family argument specifies the BINOMIAL, LOGISTIC, POISSON, GAMMA, INVERSE_GAUSSIAN, or INVERSE_BINOMIAL family, the name of the column is z_score.

The z-score is a measure of the likelihood that the NULL hypothesis is true, given this sample. It is derived by taking the difference between the observed sample mean and the hypothesized mean, divided by the standard error. The z-score statistic follows the N(0,1) distribution.

If the Family argument specifies the GAUSSIAN family, the name of the column is t_score. The t_score statistic follows a t(N-p-1) distribution.

p_value The significance level (p-value) for each predictor.
significance The likelihood that the predictor is significant (see Output in the function: CoxPH).

The output includes a row for each of the following with a value for estimated value, standard error, z-score, p-value, and significance:

GLM Output Table Parameters
Parameter Description
Loglik The log likelihood of the model.
Intercept The value of the logit (Y) when all predictors are 0.
Predictors A row for each predictor value (X1,X2,...,Xp). Each numeric input column corresponds to one predictor.