TDigest

v1.0.2

T-Digest is an online percentile and quantile estimation algorithm that can be used to estimate highly accurate values of the median, 90th percentile, 99th percentile etc of a data set while only storing a small bounded fraction of the dataset. This is useful for computing percentiles on arbitrarily large data-sets, or streaming data. This is useful because true percentiles require having the entire data-set in sorted order, which isn't feasible for many situations. Original research paper on T-Digest can be found here: https://github.com/tdunning/t-digest/blob/master/docs/t-digest-paper/histo.pdf Example usage: using StatsLib; ... TDigest digest = new TDigest(); Random r = new Random(); for (int i=0; i<1000000; i++) { digest.Add(r.NextDouble()); } var median = digest.Quantile(.5); var n90th = digest.Quantile(.9); var n99th = digest.Quantile(.99); var n999th = digest.Quantile(.999);