loess

Documentation

Overview

Example (ExternalWeights)

Output:

Smoothed [5.00 3.50 2.00 3.50 5.00 5.00 4.25 3.50 4.25 5.00 5.00 4.25 3.50 4.25 5.00 5.00 3.75 2.50 3.75 5.00]

Example (Smooth)

Output:

Smoothed [5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00]
Smoothed [5.50 4.60 3.86 3.93 4.86 5.43 5.21 4.50 4.14 4.71 5.14 4.78 4.50 4.57 5.00 5.43 4.93 4.22 4.35 4.78]

Index

• func Linear(s *State, x, left, right float64) error
• func Quadratic(s *State, x, left, right float64) error
• func Regress(s *State, fn WeightUpdate, x, left, right float64) (retVal float64, err error)
• func Smooth(x []float64, width, jump int, fn WeightUpdate) ([]float64, error)
• func UnsafeSmooth(regression *State, width, jump int, fn WeightUpdate, retVal []float64) ([]float64, error)
• type State
  • func New(width int, x []float64) *State
  • func NewWithExternal(width int, x, e []float64) *State
  • func (s State) E() []float64
  • func (s State) W() []float64
  • func (s State) X() []float64
• type WeightUpdate

Examples

• Package (ExternalWeights)
• Package (Smooth)

Functions

func Linear

func Linear(s *State, x, left, right float64) error

Linear performs linear regression, but constrained to left and right values. The implementation is slightly different from the typical Y = (X'X)^-1(X'Y)

func Quadratic

func Quadratic(s *State, x, left, right float64) error

Quadratic performs quadratic regression, constrained to left. and right.

func Regress

func Regress(s *State, fn WeightUpdate, x, left, right float64) (retVal float64, err error)

Regress performs local regression of x between left and right

func Smooth

func Smooth(x []float64, width, jump int, fn WeightUpdate) ([]float64, error)

Smooth smooths a slice of float64, with the provided width, jumps and update functions

func UnsafeSmooth

func UnsafeSmooth(regression *State, width, jump int, fn WeightUpdate, retVal []float64) ([]float64, error)

UnsafeSmooth is like Smooth, except the state is passed in, as well as a optional return value to be mutated.

Types

type State

type State struct {
	// contains filtered or unexported fields
}

State represents a state for regression. Every field will be mutated by the Regress and Smooth functions

func New

func New(width int, x []float64) *State

New creates a new LOESS state

func NewWithExternal

func NewWithExternal(width int, x, e []float64) *State

NewWithExternal creates a new LOESS state with externally provided weights

func (State) E

func (s State) E() []float64

W returns the externally defined weights in the state.

func (State) W

func (s State) W() []float64

W returns the weights in the state

func (State) X

func (s State) X() []float64

X returns the Xs in the state.

type WeightUpdate

type WeightUpdate func(s *State, x, left, right float64) error

WeightUpdate is a function that modifies the State.