dotnet/machinelearning · error · NotImplementedException

{metric} is not supported!

Error message

{metric} is not supported!

What it means

The RegressionTrialRunner's GetMetric maps RegressionMetric enum values to metric fields; only RootMeanSquaredError, RSquared, MeanSquaredError, and MeanAbsoluteError are mapped. Any other RegressionMetric value reaches the default arm and throws NotImplementedException.

Source

Thrown at src/Microsoft.ML.AutoML/API/RegressionExperiment.cs:477

                throw;
            }
        }

        public void Dispose()
        {
            _context.CancelExecution();
            _context = null;
        }

        private double GetMetric(RegressionMetric metric, RegressionMetrics metrics)
        {
            return metric switch
            {
                RegressionMetric.RootMeanSquaredError => metrics.RootMeanSquaredError,
                RegressionMetric.RSquared => metrics.RSquared,
                RegressionMetric.MeanSquaredError => metrics.MeanSquaredError,
                RegressionMetric.MeanAbsoluteError => metrics.MeanAbsoluteError,
                _ => throw new NotImplementedException($"{metric} is not supported!"),
            };
        }
    }
}

View on GitHub (pinned to 7b76e69cf9)

Solutions

1. Set the experiment's optimizer metric to RootMeanSquaredError, RSquared, MeanSquaredError, or MeanAbsoluteError.
2. Validate the configured metric against the supported set before starting the sweep.
3. Extend the switch with an arm for the required metric and upstream the change.

Example fix

// before
var exp = ctx.Auto().CreateRegressionExperiment(data);
exp.SetMetric(RegressionMetric.PoissonLoss); // throws in runner
// after
exp.SetMetric(RegressionMetric.RootMeanSquaredError);

private static readonly RegressionMetric[] SupportedMetrics =
{
    RegressionMetric.RootMeanSquaredError,
    RegressionMetric.RSquared,
    RegressionMetric.MeanSquaredError,
    RegressionMetric.MeanAbsoluteError
};
// before running:
if (!SupportedMetrics.Contains(metric)) throw new ArgumentException($"Metric {metric} not supported by the regression runner");

bool IsSupported(RegressionMetric m) => m is RegressionMetric.RootMeanSquaredError or RegressionMetric.RSquared or RegressionMetric.MeanSquaredError or RegressionMetric.MeanAbsoluteError;

try { exp.SetMetric(metric); }
catch (NotImplementedException) { exp.SetMetric(RegressionMetric.RootMeanSquaredError); // safe default
}

When it happens

Trigger: Requesting a RegressionMetric not in the four supported arms (e.g. RootMeanSquaredErrorTransformed, PoissonLoss, or LogLoss) when the runner computes the score of a trial.

Common situations: Choosing an optimizer metric that exists on the RegressionMetric enum but is not wired into the AutoML runner; enum members added in newer ML.NET versions; configs specifying a loss metric intended for a different trainer.

Related errors

• {metric} is not supported!
• {metric} is not supported!
• joinAlgorithm
• The runner metric manager is of type {_metricManager.GetType
• {fieldType.Name}

AI-assisted analysis of dotnet/machinelearning@7b76e69cf9 (2026-09-11). Data as JSON: /api/errors/5e6ef8635147c855. Report an issue: GitHub.