ResidualAnalysisOptions

Options & Configuration · AiDotNet.Evaluation.Options

Configuration options for residual analysis in regression models.

For Beginners

Residuals are errors: Actual - Predicted. Good residuals should:

  • Have mean zero (no systematic bias)
  • Be normally distributed (for inference)
  • Have constant variance (homoscedasticity)
  • Be independent (no patterns over time or by predicted value)

If these assumptions are violated, your model might be missing something.

How It Works

Residual analysis examines the differences between predicted and actual values to diagnose model problems like heteroscedasticity, non-normality, and autocorrelation.

Constructors

Constructor Summary
ResidualAnalysisOptions()

Properties

Property Summary
bool? ComputeDeletedResiduals { get; set; } Whether to compute deleted residuals.
bool? ComputePartialResidualPlots { get; set; } Whether to compute partial residual plots.
bool? ComputeResidualACF { get; set; } Whether to compute residual ACF/PACF for autocorrelation.
bool? ComputeStandardizedResiduals { get; set; } Whether to compute standardized residuals.
bool? ComputeStudentizedResiduals { get; set; } Whether to compute studentized residuals.
bool? GenerateQQPlot { get; set; } Whether to generate Q-Q plot data.
bool? GenerateResidualLeveragePlot { get; set; } Whether to generate residual vs leverage plot data.
bool? GenerateResidualVsFittedPlot { get; set; } Whether to generate residual vs fitted plot data.
bool? GenerateScaleLocationPlot { get; set; } Whether to generate scale-location plot data.
bool? IdentifyOutliers { get; set; } Whether to identify outliers.
bool? RunBreuschPaganTest { get; set; } Whether to run Breusch-Pagan test for heteroscedasticity.
bool? RunDurbinWatsonTest { get; set; } Whether to run Durbin-Watson test for autocorrelation.
bool? RunGoldfeldQuandtTest { get; set; } Whether to run Goldfeld-Quandt test.
bool? RunJarqueBeraTest { get; set; } Whether to run Jarque-Bera normality test.
bool? RunKolmogorovSmirnovTest { get; set; } Whether to run Kolmogorov-Smirnov test.
bool? RunLjungBoxTest { get; set; } Whether to run Ljung-Box test for autocorrelation.
bool? RunRESETTest { get; set; } Whether to run RESET test for functional form.
bool? RunShapiroWilkTest { get; set; } Whether to run Shapiro-Wilk normality test.
bool? RunWhiteTest { get; set; } Whether to run White test for heteroscedasticity.
double? OutlierThreshold { get; set; } Threshold for outlier detection (standard deviations).
double? SignificanceLevel { get; set; } Significance level for all tests.
int? LjungBoxLag { get; set; } Lag for Ljung-Box test.
int? MaxACFLag { get; set; } Maximum lag for ACF computation.
int? MaxOutliersToReport { get; set; } Maximum number of outliers to report.
int[] PartialResidualFeatures { get; set; } Features for partial residual plots.
int[] RESETPowers { get; set; } Power terms for RESET test.