Error codes in predictive model output

You can include an error field in the outputs of predictive models. It is useful for troubleshooting when decisions that use predictive models are executed.
The error fields contain an error code and a message.
Table 1. Error code and message
Code Message Description Example
602 MAPPING_INPUT_EXCEPTION

An error occurred when transforming the input into JSON for the query to the machine learning provider.

An exception occurred while mapping prediction input for provider {0} with deploymentId {1}

JSON processing exception: {0}
603 NULL_PREDICTION

Although the query to the machine learning provider was successful, no response was recovered.

A null prediction map was returned with deployment ID {0}, although the return was successful.
606 HTTP_RESPONSE_READING_ERROR

This error is returned when the response from the query to the machine learning provider might not be understood.

  
608 GET_PREDICTION_MAP_ERROR

The response from the query to the machine learning provider cannot be mapped to the expected output schema.

  
609 GET_PROVIDER_ERROR

This error is returned when the machine learning provider cannot be found in the runtime environment.

The provider "{0}" has not been found.

Invalid type for provider {0}.
610 NULL_RUN_CONTEXT_ERROR

The run context was not passed to the prediction call.

The run context was null.
611 GET_PROVIDER_INFO_ERROR

The providerInfo.json file was not found at prediction time or is in the wrong type.

The provider information "{0}" was not found.

The provider information "{0}" is not of the "{1}" type.