How to find periodic ripples in a data series?

Question

Analytics

Given a sequence of certain numbers, for example, the readings of a sensor. And the hypothesis that there is a cyclic pattern in these data.
We need to find the length of this cycle. Maybe every week is roughly similar to the previous ones: on Wednesday there is a peak, on Friday there is a decline, or another week-long pattern repeats itself steadily.
Or maybe not a 7-day, but say, an 11-day cycle, but repetitions of the 11-day pattern (not 1:1, but in approximation) are obvious.
Question : what is the name of this task of finding the period with the greatest “resonance” in the repetition of the pattern, and how can it be easily implemented in any PL?
I recently did something remotely similar with linear regression - there the two-dimensional data set more or less fit along a straight line on the graph. And there was a simple formula to find the coefficients from the formula describing it y = A*x + B
. Here I would like to know about the same simple formula for finding the cycle length.
Maybe someone is familiar with the mat. statistics, or signal analysis? Finding the pitch in an audio sample. Or looking for cycles in the stock market.

Answer 1

Spectral analysis to find cycles
AND autocorrelation to find patterns