Lomb-Scargle Periodogram

Description

Estimates periodicity in unevenly sampled data without interpolation, avoiding the false cycles that standard FFTs create from irregular, event-driven behavioural timestamps.

Astronomical observations are irregularly spaced, so Lomb and Scargle developed a least-squares fit of sinusoids at candidate frequencies that works directly on uneven timestamps. Behavioural data are even more event-driven: users act when motivated, not on a schedule. Applying a standard FFT produces aliasing (false periodicities from uneven intervals) and leakage (energy spread across frequencies), and treats meaningful silences such as disengagement as noise. The Behavioral Spectral Framework therefore prescribes Lomb-Scargle for irregular event sampling, alongside Gaussian-process interpolation with uncertainty bounds, survival or hazard models for inactivity intervals and point-process models, and insists that gaps be treated as behaviour (rest, hesitation, reflection) rather than void.

Attributes

Loss type
Unmigrated
Stage
Post-launch
Effort
Days

Steps

  1. Collect raw event timestamps without forcing a uniform grid.
  2. Choose a frequency grid spanning the rhythms of interest (e.g. daily, weekly, within-session).
  3. Compute the Lomb-Scargle periodogram and assess peak significance (e.g. false-alarm probability).
  4. Separately model long inactivity gaps with survival or hazard functions.
  5. Compare periodograms across users or cohorts.
  6. Interpret peaks with qualitative and contextual evidence.

Sources

  • Vivek Gopinath (2025), From Stars to Signals: Translating Matteo Paz's Fourier-Wavelet Model into Behavioral and UX Research Methodology
  • Matteo Paz (2024), A Sub-Millisecond Fourier and Wavelet Based Model to Extract Variable Candidates from the NEOWISE Single-Exposure Database

Also known as

  • Lomb-Scargle
  • Least-squares spectral analysis
  • Non-uniform periodogram

Categories

  • Information Theory, Physics & Mathematics