numpyvar

Numpyvar is not an official function in the NumPy library; it is a colloquial shorthand sometimes used to refer to computing the variance of data with NumPy. The actual, documented function is numpy.var, which returns the variance of an array or a reduced axis of an array.

The standard function is numpy.var(a, axis=None, dtype=None, out=None, ddof=0, keepdims=False). It computes the variance as the mean of the squared deviations from the mean, with options to control the axis over which the calculation is performed, the data type used for computation, and whether to keep reduced dimensions. The parameter ddof adjusts the divisor from N to N - ddof, so ddof=0 yields the population variance and ddof=1 yields the sample variance. keepdims, when True, preserves reduced dimensions in the result.

Key behavior and usage notes: if axis is None, numpy.var returns a scalar; otherwise, it returns an array with the appropriate axes reduced. Integer inputs are promoted to floating point for the computation to avoid truncation. For complex data, variance is computed in a manner that yields a real, nonnegative result, typically using the squared magnitude of deviations. Numpy.var also supports masked arrays through numpy.ma, and can be combined with numpy.mean, numpy.std, and related functions for statistical analysis.

Examples illustrate typical usage: numpy.var([1, 2, 3, 4]) evaluates to 1.25 (population variance), while numpy.var([1, 2, 3, 4], ddof=1) evaluates to 1.666..., the sample variance. See also numpy.mean and numpy.std for related statistics.