obspy.signal.trigger.aic_simple

aic_simple(a)[source]

Simple Akaike Information Criterion [Maeda1985].

It’s computed directly from input data \(a\) and defined as

\[\text{AIC}(k) = k\log(\text{Var}(a_{1..k})) + (N-k-1)\log(\text{Var}(a_{k+1..N}))\]

which variance denoted as \(\text{Var}\).

The true output is one data sample less. To make it convenient with other metrics in this module, where the output length is preserved, the last element is appended to the output: aic[-2] == aic[-1].

Parameters

a (numpy.ndarray or list) – Input time series

Return type

numpy.ndarray

Returns

aic - Akaike Information Criterion array