Cannot see confusion matrix in pipeline metrics

TensorFlow Special Interest Groups TensorBoard
,
1

I have an evaluator function and inside which I have tfma.evalConfig which contains different metrics for binary classification I can view at run time. But the only metric I cannot visualize is confusion matrix plot. What am I missing? sample code below how I set EvalConfig

                eval_config=tfma.EvalConfig(
                    model_specs=[
                        tfma.ModelSpec(
                            signature_name=serving_signature,
                            label_key="target",
                        )
                    ],
                    slicing_specs=[
                        tfma.SlicingSpec(),
                        tfma.SlicingSpec(feature_keys=["userid"]),
                    ],
                    metrics_specs=[
                        tfma.MetricsSpec(
                            metrics=[
                                tfma.MetricConfig(class_name="CalibrationPlot"),
                                tfma.MetricConfig(class_name="MeanLabel"),
                                tfma.MetricConfig(class_name="MeanPrediction"),
                                tfma.MetricConfig(class_name="ConfusionMatrixPlot"),
                            ]
                        )
                    ],
                ),
                example_splits=[split_name],
            )
3

The root cause for not being able to visualize the confusion matrix plot in your TensorFlow Model Analysis (TFMA) setup could likely be due to the absence of the required plot_config in your MetricConfig for the ConfusionMatrixPlot . Ensure that you include a plot_config with the necessary parameters specified for the confusion matrix plot within your MetricConfig .