EXPLAIN_MODEL function
The EXPLAIN_MODEL function returns a SUPER data type that contains a model explainability report in a JSON format. The explainability report contains information about the Shapley value for all model features.
The EXPLAIN_MODEL function currently supports only the AUTO ON or AUTO OFF XGBoost models.
When the explainability report isn't available, the function returns statuses
showing on the progress of the model. These include Waiting for training job to
complete, Waiting for processing job to complete, and
Processing job failed.
When you run the CREATE MODEL statement, the explanation state becomes Waiting
for training job to complete. When the model has been trained and an
explanation request is sent, the explanation state becomes Waiting for processing
job to complete. When the model explanation completes successfully, the full
explainability report is available. Otherwise, the state becomes Processing job
failed.
When you run the CREATE MODEL statement, you can use the optional MAX_RUNTIME parameter
to specify the maximum amount of time the training should take. Once model creation reaches that amount of time,
Amazon Redshift stops creating the model. If you reach that time limit while creating an autopilot
model, Amazon Redshift will return the best model so far. Model explainability becomes available once the model
training finishes, so if MAX_RUNTIME is set to a low amount of time, the explainability report might
not be available. Training time varies and depends on
model complexity, data size, and other factors.
Syntax
EXPLAIN_MODEL ('schema_name.model_name')
Argument
- schema_name
-
The name of the schema. If no schema_name is specified, then the current schema is selected.
- model_name
-
The name of the model. The model name in a schema must be unique.
Return type
The EXPLAIN_MODEL function returns a SUPER data type, as shown following.
{"version":"1.0","explanations":{"kernel_shap":{"label0":{"global_shap_values":{"x0":0.05,"x1":0.10,"x2":0.30,"x3":0.15},"expected_value":0.50}}}}
Examples
The following example returns the explanation state waiting for training job to complete.
select explain_model('customer_churn_auto_model'); explain_model -------------------------------------------------------- {"explanations":"waiting for training job to complete"} (1 row)
When the model explanation completes successfully, the full explainability report is available as follows.
select explain_model('customer_churn_auto_model'); explain_model ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ {"version":"1.0","explanations":{"kernel_shap":{"label0":{"global_shap_values":{"x0":0.05386043365892927,"x1":0.10801289723274592,"x2":0.23227865827017378,"x3":0.0676685133940455,"x4":0.0897097667672375,"x5":0.08502141653270926,"x6":0.07581993936077065,"x7":0.16462880604578135},"expected_value":0.8492974042892456}}}} (1 row)
Because the EXPLAIN_MODEL function returns the SUPER data type, you can query the
explainability report. By doing this, you can extract
global_shap_values, expected_value, or feature-specific
Shapley values.
The following example extracts global_shap_values for the model.
select json_table.report.explanations.kernel_shap.label0.global_shap_values from (select explain_model('customer_churn_auto_model') as report) as json_table; global_shap_values -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- {"state":0.10983770427197151,"account_length":0.1772441398408543,"area_code":0.08626823968639591,"phone":0.0736669595282712,"intl_plan":3.344907436910987,"vmail_plan":0.09646600597854467,"vmail_message":0.2064922655089351,"day_mins":2.015038015251777,"day_calls":0.13179511076780168,"day_charge":0.4941091720480879,"eve_mins":0.46081379198626105,"eve_calls":0.16913440417758477,"eve_charge":0.09651014369401761,"night_mins":0.44218153640050845,"night_calls":0.15311640089218997,"night_charge":0.13850366104495426,"intl_mins":0.7583662464883899,"intl_calls":0.47144468610485685,"intl_charge":0.10945894673611875,"cust_serv_calls":0.31822051038387733} (1 row)
The following example extracts global_shap_values for the feature
x0.
select json_table.report.explanations.kernel_shap.label0.global_shap_values.x0 from (select explain_model('customer_churn_auto_model') as report) as json_table; x0 ------------------------ 0.05386043365892927 (1 row)
If the model is created in a specific schema and you have access to the created model, then you can query the model explanation as shown following.
-- Check the current schema SHOW search_path; search_path ------------------ $user, public (1 row) -- If you have the privilege to access the model explanation -- in `test_schema` SELECT explain_model('test_schema.test_model_name'); explain_model --------------------------------------------------------- {"explanations":"waiting for training job to complete"} (1 row)