nestedsoftware/iterative_stats

Demo of adjustment to Welford method for calculating mean/variance/stdev that incorporates a sliding window of fixed size over the incoming data

iterative_stats's Issues

Fails very basic tests, when dSquared approaches zero

It failed a very basic test, comparing a few integer inputs against the brute force calculations, even though I was only looking for accuracy within two decimal places.

I tried using these inputs with a buffer size of 6, it will fail on update(samples[19]) when all six values are now 6:

const samples [ 1, 2, 3,3,3,
4,4,4,4,
5,5,5,5,5,
6,6,6,6,6,6
6,6,6,6,6]

I "fixed" it by resetting the value of dSquared to actual zero when "too close" to zero
if(Math.abs(this._dSquared) < 0.000_000_000_001) { this._dSquared = 0 }