An extension of OnlineStatsBase.jl that supports proper statistical weighting
and arbitrary numerical precision.
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)WeightedOnlineStats.jl currently implements the following Statistics:
WeightedSumWeightedMeanWeightedVarianceWeightedCovMatrixWeightedHistWeightedAdaptiveHist