TheAlgorithms/C-Sharp · error · InvalidOperationException

No training data available.

Error message

No training data available.

What it means

Predict cannot classify without any training samples: there is nothing to compare distances against. The library throws InvalidOperationException when trainingData.Count == 0, i.e. Predict was called before any AddSample call.

Solutions

1. Call AddSample at least once (with valid k >= 1 data) before calling Predict.
2. Verify the training data source actually loaded samples; check count > 0 after loading.
3. Expose a check like trainingCount > 0 in application code before invoking Predict.

Example fix

// before
var knn = new KNearestNeighbors(3);
var label = knn.Predict(query); // no samples added
// after
var knn = new KNearestNeighbors(3);
foreach (var s in trainingSet) knn.AddSample(s.Features, s.Label);
if (trainingSet.Count == 0) throw new InvalidOperationException("Training set is empty.");
var label = knn.Predict(query);

if (loadedSamples.Count == 0)
    throw new InvalidOperationException("Cannot predict: no training samples were loaded.");
var label = knn.Predict(query);

static bool IsReadyForPrediction<TLabel>(KNearestNeighbors<TLabel> knn) => knn != null && loadedSampleCount > 0;

try
{
    var label = knn.Predict(query);
}
catch (InvalidOperationException)
{
    // model has no training data — train first or return a 'model not ready' response
    throw new ApplicationException("Classifier not trained yet.");
}

When it happens

Trigger: Calling Predict on a freshly constructed KNearestNeighbors instance with no prior AddSample calls; all AddSample calls skipped due to an empty or filtered training file.

Common situations: Training data file empty or path wrong so no samples loaded; loading code silently skipped malformed rows; model constructed but the training step accidentally removed during refactoring; deserialized model losing its training data.

Understand the failure class

Background: "Invalid state transition" errors: "status must be X, actually Y", "already rejected/charging/uninstalled", "cannot ... while running" — what they mean when a library rejects your call — this error's family across 31 libraries.

Related errors

• Model must be fitted before prediction.
• k must be at least 1.
• Feature vectors must be of the same length.
• ArgumentNullException: features
• Input data cannot be null.

AI-assisted analysis of TheAlgorithms/C-Sharp@96e2905cab (2026-09-13). Data as JSON: /api/errors/18a3a69ec82c5c69. Report an issue: GitHub.

Appendix: source

Thrown at Algorithms/MachineLearning/KNearestNeighbors.cs:83

        if (features == null)
        {
            throw new ArgumentNullException(nameof(features));
        }

        trainingData.Add((features, label));
    }

    /// <summary>
    /// Predicts the label for a given feature vector using the KNN algorithm.
    /// </summary>
    /// <param name="features">Feature vector to classify.</param>
    /// <returns>Predicted label.</returns>
    /// <exception cref="InvalidOperationException">Thrown if there is no training data.</exception>
    public TLabel Predict(double[] features)
    {
        if (trainingData.Count == 0)
        {
            throw new InvalidOperationException("No training data available.");
        }

        if (features == null)
        {
            throw new ArgumentNullException(nameof(features));
        }

        // Compute distances to all training samples
        var distances = trainingData
            .Select(td => (Label: td.Label, Distance: EuclideanDistance(features, td.Features)))
            .OrderBy(x => x.Distance)
            .Take(k)
            .ToList();

        // Majority vote
        var labelCounts = distances
            .GroupBy(x => x.Label)
            .Select(g => new { Label = g.Key, Count = g.Count(), MinDistance = g.Min(item => item.Distance) })

View on GitHub (pinned to 96e2905cab)