I am trying to use parquet as the data format in an AWS Sagemaker.

But when trying to use parquet as the dataformat for AWS ClarifyCheckStep in a pipeline, I will get ClientError: Content type is missing when using Parquet dataset. Please specify the content type

sample code looks like this

from sagemaker.clarify import DataConfig
from sagemaker.workflow.clarify_check_step import (
    ClarifyCheckStep,
    ModelExplainabilityCheckConfig,
    SHAPConfig
)

output_model_explainability = "s3://foo/bar"
model_explainability_analysis_cfg_output_path = "s3://foo/baz"
config_model_explainability_data = DataConfig(
    s3_data_input_path=step_preprocess.properties.ProcessingOutputConfig.Outputs[
        "train"
    ].S3Output.S3Uri,
    s3_output_path=output_model_explainability,
    s3_analysis_config_output_path=model_explainability_analysis_cfg_output_path,
    label=0,
    dataset_type="application/x-parquet",
)
config_shap = SHAPConfig(seed=42, num_samples=10)
config_check_model_explainability = ModelExplainabilityCheckConfig(
    data_config=config_model_explainability_data,
    model_config=config_model,
    explainability_config=config_shap,
)
step_check_model_explainability = ClarifyCheckStep(
    name="ModelExplainabilityCheckStep",
    clarify_check_config=config_check_model_explainability,
    check_job_config=config_check_job,
    skip_check=True,
    register_new_baseline=True,
    supplied_baseline_constraints=True,
    model_package_group_name="packagename",
)