Least-Squares Spectral Analysis

Least-squares spectral analysis is a class of methods for estimating the frequency spectrum of a dataset by fitting sinusoids to the data using a least-squares fit, rather than through the Fourier transform. Petr Vanicek developed the first strictly least-squares version between 1969 and 1971, and Nicholas Lomb and Jeffrey Scargle later developed the related Lomb-Scargle periodogram in the 1970s and 1980s. Unlike Fourier analysis, it does not require evenly spaced data points, so it can be applied directly to incomplete or unevenly sampled time series without first altering the data. Because Fourier analysis tends to amplify long-period noise in long or gapped records, least-squares spectral analysis is often preferred for such data, with spectral magnitudes showing how much each frequency contributes to the variance of the series. This description is adapted from Wikipedia contributors under CC BY-SA 4.0; changes were made. https://creativecommons.org/licenses/by-sa/4.0/