DatasetStatistics<T>

Models & Types · AiDotNet.Evaluation.Results.Core

Statistics about the dataset used for evaluation.

For Beginners

Before trusting evaluation metrics, you need to understand your data. This class tells you:

  • How many samples were evaluated
  • Class distribution (for classification) - are classes balanced?
  • Target variable statistics (for regression) - range and spread
  • Missing values, outliers, and other data quality issues

How It Works

Contains summary information about the evaluation dataset including sample counts, class distributions, feature statistics, and data quality indicators.

Constructors

Constructor Summary
DatasetStatistics()

Properties

Property Summary
IReadOnlyDictionary<T, double> ClassProportions { get; set; } Proportion of samples per class.
IReadOnlyDictionary<T, int> ClassCounts { get; set; } Number of samples per class.
IReadOnlyList<FeatureStatistics<T>> FeatureStats { get; set; } Per-feature statistics (min, max, mean, std, missing count).
IReadOnlyList<T> ClassLabels { get; set; } Class labels (for classification).
IReadOnlyList<string> FeatureNames { get; set; } Feature names, if available.
IReadOnlyList<string> Warnings { get; set; } Warnings about the dataset (imbalance, outliers, etc.).
T TargetMax { get; set; } Maximum target value (for regression).
T TargetMean { get; set; } Mean target value (for regression).
T TargetMedian { get; set; } Median target value (for regression).
T TargetMin { get; set; } Minimum target value (for regression).
T TargetStdDev { get; set; } Standard deviation of target (for regression).
bool AppearsTimeSeries { get; set; } Whether the dataset appears to be a time series (ordered, potentially autocorrelated).
bool IsClassification { get; set; } Whether this is a classification task.
bool IsMultiLabel { get; set; } Whether this is a multi-label classification task.
bool? IsImbalanced { get; set; } Whether classes are considered imbalanced (ratio > threshold).
double MissingProportion { get; } Proportion of samples with missing values.
double OutlierProportion { get; } Proportion of outliers.
double? ClassImbalanceRatio { get; set; } Class imbalance ratio (largest class / smallest class).
double? TargetKurtosis { get; set; } Target value kurtosis (for regression).
double? TargetSkewness { get; set; } Target value skewness (for regression).
int DuplicateCount { get; set; } Number of duplicate samples.
int MissingSamplesCount { get; set; } Number of samples with missing values.
int NumberOfFeatures { get; set; } Number of features in the dataset.
int NumberOfOutputs { get; set; } Number of output dimensions (1 for scalar, >1 for multi-output).
int OutlierCount { get; set; } Number of detected outliers.
int TotalSamples { get; set; } Total number of samples in the evaluation set.
int? NumberOfClasses { get; set; } Number of unique classes (for classification).

Methods

Method Summary
string GetSummary() Gets a summary string describing the dataset.