IncrementalMoments

Defined in fynance.features.engineering

class IncrementalMoments

Bases: object

Online mean / variance via Welford’s algorithm (O(1) per update).

Streaming alternative to recomputing a rolling mean/variance from scratch.

Attributes:

nint

Number of observations seen.

meanfloat

Running mean.

Examples

>>> im = IncrementalMoments()
>>> for v in [1.0, 2.0, 3.0]:
...     _ = im.update(v)
>>> im.mean, round(im.var, 4)
(2.0, 0.6667)

property std

Population standard deviation.

update(x)

Incorporate one observation; return self for chaining.

property var

Population variance (0 before the second observation).