adwinv

README

ADWIN-V

ADWIN-V is an adaptive windowing algorithm for vector data. It detects a change point using the vector magnitude and cosine similarity between mean vector and the vector. The change detection algorithm is ADWIN or ADWIN2 which is from Learning from time-changing data with adaptive windowing, Bifet, Albert, and Ricard Gavalda; Proceedings of the 2007 SIAM international conference on data mining. Society for Industrial and Applied Mathematics, 2007.

See also:

• https://www.cs.upc.edu/~gavalda/DataStreamSeminar/files/Lecture6.pdf

Usage

ADWIN-V

ADWIN-V uses ADWIN for change detection of vector magnitude and cosine similarity.

        dim := 2
	deltaM := 0.01
	deltaA := 0.01
	adwinv := NewAdwinV(dim, deltaM, deltaA)
	adwinv.Conservative(true) // if you wants
	adwinv.ScaleMagnitudes(0.1) // if you wants
	adwinv.ScaleAngles(1.0) // if you wants

	// Add stream data
	adwinv.Add([]float64{1.0, 2.0})

	// Show status
	adwinv.Size()
	adwinv.SizeMagnitudes()
	adwinv.SizeAngles()
	adwinv.Mean()
	adwinv.Detected()

ADWIN2-V

ADWIN2-V uses ADWIN2 for change detection of vector magnitude and cosine similarity.

• Note: This version provides SyncWindow option to synchronize the bucket size of the exponential histogram.

        dim := 2
	deltaM := 0.01
	deltaA := 0.01
	adwinv := NewAdwin2V(dim, deltaM, deltaA)
	adwinv.SyncWindow(true) // if you wants
	adwinv.Conservative(true) // if you wants
	adwinv.ScaleMagnitudes(0.1) // if you wants
	adwinv.ScaleAngles(1.0) // if you wants

	// Add stream data
	adwinv.Add([]float64{1.0, 2.0})

	// Show status
	adwinv.Size()
	adwinv.SizeMagnitudes()
	adwinv.SizeAngles()
	adwinv.Mean()
	adwinv.Detected()

Installation

$ go get github.com/monochromegane/adwin-v

License

MIT

Author

monochromegane

Documentation

Index

• type AdwinV
  • func NewAdwin2V(dim int, deltaM, deltaA float64) *AdwinV
  • func NewAdwinV(dim int, deltaM, deltaA float64) *AdwinV
  • func (a *AdwinV) Add(x []float64)
  • func (a *AdwinV) Conservative(t bool)
  • func (a *AdwinV) Detected() bool
  • func (a *AdwinV) Mean() []float64
  • func (a *AdwinV) ScaleAngles(scale float64)
  • func (a *AdwinV) ScaleMagnitudes(scale float64)
  • func (a *AdwinV) Size() int
  • func (a *AdwinV) SizeAngles() int
  • func (a *AdwinV) SizeMagnitudes() int
  • func (a *AdwinV) SyncWindow(t bool)

Types

type AdwinV

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

func NewAdwin2V

func NewAdwin2V(dim int, deltaM, deltaA float64) *AdwinV

func NewAdwinV

func NewAdwinV(dim int, deltaM, deltaA float64) *AdwinV

func (*AdwinV) Add

func (a *AdwinV) Add(x []float64)

func (*AdwinV) Conservative

func (a *AdwinV) Conservative(t bool)

func (*AdwinV) Detected

func (a *AdwinV) Detected() bool

func (*AdwinV) Mean

func (a *AdwinV) Mean() []float64

func (*AdwinV) ScaleAngles

func (a *AdwinV) ScaleAngles(scale float64)

func (*AdwinV) ScaleMagnitudes

func (a *AdwinV) ScaleMagnitudes(scale float64)

func (*AdwinV) Size

func (a *AdwinV) Size() int

func (*AdwinV) SizeAngles

func (a *AdwinV) SizeAngles() int

func (*AdwinV) SizeMagnitudes

func (a *AdwinV) SizeMagnitudes() int

func (*AdwinV) SyncWindow

func (a *AdwinV) SyncWindow(t bool)