OnlineStatistics

A streaming one-pass accumulator for descriptive statistics.

Computes mean, variance, standard deviation, skewness, and kurtosis without storing individual data points. Uses the Terriberry (2008) extension of Welford's online algorithm for numerically stable single-pass updates of central moments M2, M3, and M4.

This is analogous to Apache Commons Math SummaryStatistics.

Non-finite values: NaN values propagate through all statistics (IEEE 754 semantics). Infinite values are handled correctly for mean, sum, min, and max (which use compensated summation). However, variance, standardDeviation, skewness, and kurtosis return NaN when infinite values are present, since the Welford algorithm used for higher moments computes differences of running means, producing NaN from Inf - Inf.

Example:

val stats = OnlineStatistics()
stats.addAll(doubleArrayOf(1.0, 2.0, 3.0, 4.0, 5.0))
stats.mean      // 3.0
stats.variance() // 2.5
stats.count      // 5

See also