predict_table#

View page source

The Predict Table: Average Integrand Predictions#

See Also#

avgint_table

Purpose#

This table contains model predictions for the average integrand .

Avgint Subset#

For each row in the predict table, the avgint_id column below identifies the corresponding row in the avgint_table . Only a subset of the rows of the avgint table is included. The subset is defined by the following node and covariate conditions:

Node#

The node must be the Parent Node , or a Descendant of the parent node.

Covariates#

All of the Covariates must satisfy the max_difference criteria.

predict_id#

This column has type integer and is the primary key for this table. Its initial value is zero, and it increments by one for each row.

sample_index#

This column has type integer and specifies the set of model_variables that the avg_integrand corresponds to. If source is fit_var or truth_var , sample_index is null (and the corresponding model variables are used). Otherwise source is sample and sample_index is the corresponding value in the sample_table . This column is monotone non-decreasing; i.e. the value in each row is greater than or equal the value in the previous row. Each sample_index value in the sample table appears multiple times in the predict table, once for each avgint_id in the of the Avgint Subset .

avgint_id#

This column has type integer and specifies the avgint_id that avg_integrand corresponds to. For each sample_index value, avgint_id is monotone increasing and includes every element of the Avgint Subset .

avg_integrand#

This column type real and is the average integrand \(A_i(u, \theta)\). The model variables \((u, \theta)\) corresponding to the model variables with sample_index in sample_table . The subscript \(i\) denotes the information in the avgint table for the specified avgint_id .

Example#

The predict_command.py is an example that creates this table.