Loess.Interface.SmoothParms

commons-math-interpolation / Loess / SmoothParms

Interface: SmoothParms

Defined in: Loess.ts:82

Parameters for smooth.

Extended by

• LoessInterpolatorParms

Properties

accuracy?

optional accuracy?: number;

Defined in: Loess.ts:124

The accuracy threshold. Must be finite and >= 0. If the median residual is less than or equal to this value, no more robustness iterations are performed. Additionally, if the weighted standard deviation of the x values within a local regression is less than this value, the slope of the local regression line is assumed to be 0.

Default Value

1E-12

---

bandwidthFraction?

optional bandwidthFraction?: number;

Defined in: Loess.ts:108

The fraction of the points (with a non-zero weight) that is used for each local regression. Must be greater than 0 and not greater than 1. The resulting number of points is limited to at least 2 and at most all points.

Default Value

0.3

---

diagInfo?

optional diagInfo?: SmoothDiagInfo;

Defined in: Loess.ts:140

An optional object to receive diagnostics info.

---

outlierDistanceFactor?

optional outlierDistanceFactor?: number;

Defined in: Loess.ts:135

The outlier distance, relative to the median residual. Must be finite and > 0. In robustness iterations, points with a residual of at least outlierDistanceFactor * medianResidual get the robustness weight 0. The other points are weighted with the bisquare function (1 - (residual / (outlierDistanceFactor * medianResidual))^2)^2. The median residual is computed over the points with a non-zero weight and a non-NaN smoothed value. Points with a NaN smoothed value get the robustness weight 0.

Default Value

6

---

robustnessIters?

optional robustnessIters?: number;

Defined in: Loess.ts:115

The maximum number of additional robustness iterations. 0 for a single regression pass without robustness weighting. Must be an integer >= 0.

Default Value

2

---

weights?

optional weights?: ArrayLike<number>;

Defined in: Loess.ts:100

Optional weights of the points. The values must be finite and should not be negative. Points with weight 0 are ignored for the local regressions, but a smoothed y value is computed for them. If undefined, all points have the weight 1.

---

xVals

xVals: ArrayLike<number>;

Defined in: Loess.ts:88

The x values of the points, in monotonically increasing order. Equal values are allowed. The values must be finite.

---

yVals

yVals: ArrayLike<number>;

Defined in: Loess.ts:93

The y values of the points. The values must be finite.