autocorrelation
UK[ˌɔːtəʊˌkɒrɪˈleɪʃən]US[ˌɔːtoʊˌkɔːrɪˈleɪʃən]
n
A mathematical measure of the similarity between a signal and a time-shifted version of itself, used to find repeating patterns.
n
In statistics, the correlation of a variable with itself over successive time intervals.
Morpheme Breakdown
auto
correlation
auto
self
correlation
mutual relationship
Etymology
The word is a modern scientific compound, formed in English by combining two classical elements. The prefix 'auto-', derived from the Greek 'autos' (self), was adopted into scientific vocabulary to denote self-operation or internal reference. The base word 'correlation' originates from the Medieval Latin 'correlationem', built from 'com-' (together) and 'relatio' (a carrying back), which itself evolved from the Latin verb 'referre'. Literally, 'autocorrelation' means the act of carrying back a relationship to itself. This elegantly captures the core concept in mathematics and statistics: measuring how a data series or signal is related to a delayed copy of itself, thereby revealing internal periodic structures or dependencies over time.
Analysis
Structure: auto (self) + correlation (mutual relationship)
- auto: From Greek 'autos' meaning "self." In English, it functions as a combining form meaning "self-acting" or "of or by oneself."
- correlation: From Latin 'com-' (together) + 'relatio' (a bringing back, report). In English, it functions as a noun meaning a mutual relationship or connection between two or more things.
Examples
The autocorrelation function is crucial for analyzing the periodicity in time-series data.
A high degree of autocorrelation in the residuals suggests the model fails to capture a pattern.
Signal processing engineers use autocorrelation to detect faint repetitive signals in noise.