ml_model_summary
Use the ml_model_summary operator to retrieve the metadata about a machine learning model that was previously trained and saved using the fit operator, presenting a clear, more interpretable summary of the model parameters, including the model features and their associated coefficients.
Specifically, the summary now displays each feature alongside its coefficient, standard error, and p-value (where applicable) in a structured table.
This summary, therefore, helps users in critical model analysis and understanding of the model output and effectiveness.
Syntax
|ml_model_summary <model_name> [display_as_table]
<model_name>
|
Name of the model previously trained and saved using the |
display_as_table
|
Choose whether to display the summary as a table. Possible values are |
Usage
Use the ml_model_summary operator to to retrieve the metadata about the machine learning model which has been trained and saved using the fit operator. The metadata it retrieves are the model algorithm, model parameters, and the model summary. The model summary fields vary for each algorithm. It might include model coefficients, intercepts, offsets, cluster details, and so on, depending on the model.
The summary table displays each feature alongside its coefficient, standard error, and p-value (where applicable), enabling you to:
-
Identify the influence of each feature on the model outcome.
-
Conduct statistical analysis or adjustments through p-values and standard errors.
For example, you can:-
Remove insignificant features.
For example, if a feature has a high p-value (for example, greater than 0.5), its contribution might be insignificant. -
Check for multicollinearity.
Consider removing or combining features that are highly correlated. -
Refit the model.
If you remove or adjust some features, consider refitting the model.
-
-
View results in a well-formatted table, simplifying model analysis.
Examples
Example 1: Retrieve model summary displayed as a formatted table.
|ml_model_summary my_model display_as_table=Tru