InfluenceAnalysisOptions
Options & Configuration · AiDotNet.Evaluation.Options
Configuration options for influence analysis in regression models.
For Beginners
Some data points have more influence on the model than others:
- High leverage: Point has unusual X values (far from center)
- High influence: Removing this point significantly changes the model
A point can be high leverage without being influential (if it fits the pattern). Influential points deserve extra scrutiny - they might be errors or key insights.
How It Works
Influence analysis identifies data points that have outsized impact on model predictions or parameter estimates. These influential points may be outliers, errors, or important cases.
Constructors
| Constructor | Summary |
|---|---|
InfluenceAnalysisOptions() |
Properties
| Property | Summary |
|---|---|
bool? ComputeApproximateLOOImpact { get; set; } |
Whether to compute approximate leave-one-out impact. |
bool? ComputeCooksDistance { get; set; } |
Whether to compute Cook’s distance. |
bool? ComputeCovarianceRatio { get; set; } |
Whether to compute covariance ratio. |
bool? ComputeDFBETAS { get; set; } |
Whether to compute DFBETAS. |
bool? ComputeDFFITS { get; set; } |
Whether to compute DFFITS. |
bool? ComputeExactLOOImpact { get; set; } |
Whether to compute exact leave-one-out (refit for each point). |
bool? ComputeHadisMeasure { get; set; } |
Whether to compute Hadi’s influence measure. |
bool? ComputeLeverage { get; set; } |
Whether to compute leverage (hat values). |
bool? ComputeWelschKuhDistance { get; set; } |
Whether to compute Welsch-Kuh distance. |
bool? GenerateAddedVariablePlots { get; set; } |
Whether to generate added variable plots. |
bool? GenerateCPRPlots { get; set; } |
Whether to generate component-plus-residual plots. |
bool? GenerateInfluencePlot { get; set; } |
Whether to generate influence plot data. |
bool? IdentifyInfluentialPoints { get; set; } |
Whether to identify influential points. |
bool? IncludeRecommendations { get; set; } |
Whether to include recommendations for handling influential points. |
bool? ReportPredictionInfluence { get; set; } |
Whether to report influence on specific predictions. |
double? CooksDistanceThreshold { get; set; } |
Cook’s distance threshold for flagging. |
double? DFBETASThreshold { get; set; } |
DFBETAS threshold for flagging. |
double? DFFITSThreshold { get; set; } |
DFFITS threshold for flagging. |
double? LeverageThreshold { get; set; } |
Leverage threshold for flagging. |
int? MaxInfluentialPointsToReport { get; set; } |
Maximum number of influential points to report. |
int[] AddedVariableFeatures { get; set; } |
Features for added variable plots. |
int[] DFBETASCoefficients { get; set; } |
Specific coefficients to compute DFBETAS for. |
int[] ExactLOOSubset { get; set; } |
Subset of points for exact LOO (indices). |
int[] PredictionIndicesToAnalyze { get; set; } |
Prediction indices to analyze influence for. |