EvaluationOptions<T>

Options & Configuration · AiDotNet.Evaluation.Options

Configuration options for model evaluation.

For Beginners

These options control how your model is evaluated. By default, the framework auto-detects what kind of model you have (classification, regression, etc.) and computes appropriate metrics. You can customize this if needed.

How It Works

Controls which metrics are computed, confidence interval settings, and output preferences. All properties are nullable with sensible defaults applied internally.

Constructors

Constructor Summary
EvaluationOptions()

Properties

Property Summary
AveragingMethod? MultiClassAveraging { get; set; } Averaging method for multi-class metrics.
ConfidenceIntervalMethod? ConfidenceIntervalMethod { get; set; } Method for computing confidence intervals.
IReadOnlyList<string> MetricsToCompute { get; set; } Which metrics to compute.
T ClassificationThreshold { get; set; } Custom classification threshold.
T FalseNegativeCost { get; set; } Cost of false negatives for cost-sensitive evaluation.
T FalsePositiveCost { get; set; } Cost of false positives for cost-sensitive evaluation.
T PositiveClassLabel { get; set; } Positive class label for binary classification.
ThresholdSelectionMethod? ThresholdSelectionMethod { get; set; } Threshold selection method for binary classification.
bool? ComputeAllMetrics { get; set; } Whether to compute all available metrics.
bool? ComputeCalibrationMetrics { get; set; } Whether to compute probability calibration metrics.
bool? ComputeConfidenceIntervals { get; set; } Whether to compute confidence intervals for metrics.
bool? ComputePermutationImportance { get; set; } Whether to compute feature importance via permutation.
bool? ComputeThresholdAnalysis { get; set; } Whether to compute threshold analysis for binary classification.
bool? EmitWarnings { get; set; } Whether to warn about potential issues (class imbalance, etc.).
bool? IncludePerClassMetrics { get; set; } Whether to include per-class metrics for classification.
bool? PerformInfluenceAnalysis { get; set; } Whether to perform influence analysis for regression.
bool? PerformResidualAnalysis { get; set; } Whether to perform residual analysis for regression.
bool? TrackComputationTime { get; set; } Whether to track computation time for each metric.
double? ConfidenceLevel { get; set; } Confidence level for intervals.
int? BootstrapSamples { get; set; } Number of bootstrap samples for confidence intervals.
int? MaxDegreeOfParallelism { get; set; } Number of parallel threads for computation.
int? PermutationRounds { get; set; } Number of permutation rounds for importance.
int? RandomSeed { get; set; } Random seed for reproducible confidence intervals.
long? MemoryLimitBytes { get; set; } Memory limit in bytes for evaluation.