API Reference
fastloess / OnlineSmoothOptions
Interface: OnlineSmoothOptions
Section titled “Interface: OnlineSmoothOptions”Defined in: index.d.ts:190
Configuration options for online LOESS smoothing.
A subset of [SmoothOptions]: diagnostics, residuals, parallel execution,
and cross-validation are all no-ops for online processing (it handles one
point at a time, always runs sequentially, and always returns a residual
inline), so they aren’t fields on this type. intervals and the “se” output
require update_mode: "full".
Properties
Section titled “Properties”auto_converge?
Section titled “auto_converge?”
optionalauto_converge?:number
Defined in: index.d.ts:209
Auto-convergence tolerance. Default: None.
boundary_degree_fallback?
Section titled “boundary_degree_fallback?”
optionalboundary_degree_fallback?:boolean
Defined in: index.d.ts:228
Reduce polynomial degree to linear at boundary vertices (default true).
boundary_policy?
Section titled “boundary_policy?”
optionalboundary_policy?:string
Defined in: index.d.ts:205
Boundary handling (“extend”, “reflect”, “zero”, “noboundary”). Default: “extend”.
optionalcell?:number
Defined in: index.d.ts:224
Interpolation cell size (default 0.2). Smaller = more vertices, higher accuracy.
degree?
Section titled “degree?”
optionaldegree?:string
Defined in: index.d.ts:214
Polynomial degree (“constant”, “linear”, “quadratic”, etc.). Default: “linear”.
dimensions?
Section titled “dimensions?”
optionaldimensions?:number
Defined in: index.d.ts:216
Number of predictor dimensions; Online vector updates accept one coordinate per dimension.
distance_metric?
Section titled “distance_metric?”
optionaldistance_metric?:string
Defined in: index.d.ts:218
Distance metric (“normalized”, “euclidean”, “manhattan”, “chebyshev”, “minkowski:p”, “weighted”). Default: “normalized”.
fraction?
Section titled “fraction?”
optionalfraction?:number
Defined in: index.d.ts:192
Smoothing fraction (0 < fraction <= 1). Default: 0.67.
interpolation_vertices?
Section titled “interpolation_vertices?”
optionalinterpolation_vertices?:number
Defined in: index.d.ts:226
Maximum number of interpolation vertices.
intervals?
Section titled “intervals?”
optionalintervals?:IntervalsOptions
Defined in: index.d.ts:212
iterations?
Section titled “iterations?”
optionaliterations?:number
Defined in: index.d.ts:197
Number of robustness iterations. Default: 0; positive values require
update_mode: "full".
missing?
Section titled “missing?”
optionalmissing?:string
Defined in: index.d.ts:230
Policy for non-finite (NaN/Inf) x/y values passed to addPoint (“error”, “drop”). Default: “error”.
outputs?
Section titled “outputs?”
optionaloutputs?:string[]
Defined in: index.d.ts:211
Optional output components: weights, gradient (or derivative), se.
robustness_method?
Section titled “robustness_method?”
optionalrobustness_method?:string
Defined in: index.d.ts:201
Robustness method (“bisquare”, “huber”, “talwar”). Default: “bisquare”.
scaling_method?
Section titled “scaling_method?”
optionalscaling_method?:string
Defined in: index.d.ts:207
Scaling method (“mad”, “mar”, “mean”). Default: “mad”.
surface_mode?
Section titled “surface_mode?”
optionalsurface_mode?:string
Defined in: index.d.ts:222
Surface mode (“interpolation” or “direct”). Default: “interpolation”.
weight_function?
Section titled “weight_function?”
optionalweight_function?:string
Defined in: index.d.ts:199
Weight function (“tricube”, “epanechnikov”, “gaussian”, “uniform”, “biweight”, “triangle”, “cosine”). Default: “tricube”.
weighted_metric_weights?
Section titled “weighted_metric_weights?”
optionalweighted_metric_weights?:number[]
Defined in: index.d.ts:220
Per-dimension weights for the “weighted” distance metric.
zero_weight_fallback?
Section titled “zero_weight_fallback?”
optionalzero_weight_fallback?:string
Defined in: index.d.ts:203
Fallback strategy when weights are zero (“use_local_mean”, “return_original”, “return_none”). Default: “use_local_mean”.