WeightedOnlineStats.jl

Weighted version of OnlineStats.jl
Author gdkrmr
Popularity
8 Stars
Updated Last
2 Years Ago
Started In
November 2018

WeightedOnlineStats.jl

Run tests Build Status DOI codecov.io

An extension of OnlineStatsBase.jl that supports proper statistical weighting and arbitrary numerical precision.

Usage

using WeightedOnlineStats

values = rand(100)
weights = rand(100)

# fit using arrays:
o1 = fit!(WeightedMean(), values, weights)

# fit using an iterator that returns a tuple (value, weight):
o2 = fit!(WeightedMean(), zip(values, weights))

# fit single values at a time:
o3 = WeightedMean()
for i in 1:length(values)
    fit!(o3, values[i], weights[i])
end

mean(o1)
mean(o2)
mean(o3)

Statistics

WeightedOnlineStats.jl currently implements the following Statistics:

  • WeightedSum
  • WeightedMean
  • WeightedVariance
  • WeightedCovMatrix
  • WeightedHist
  • WeightedAdaptiveHist

Used By Packages