Fit a sine curve using non-linear least squares — fit_sine

This function fits a sine curve of the form \(y = \text{amp} \cdot \sin\left(\frac{2\pi}{\text{period}} \cdot (x + \text{phase})\right) + \text{offset}\) to time series data using Levenberg-Marquardt optimization from `minpack.lm::nls.lm`.

Usage

fit_sine_curve(
  x,
  y,
  parStart = list(amp = 3, phase = 0, offset = 0, period = 24)
)

Arguments

x: Numeric vector of time values (e.g., hours or days).
y: Numeric vector of response values.
parStart: Named list with starting values for parameters: `amp`, `phase`, `offset`, and `period`.

Value

A named list with:

amp: Amplitude (peak distance from baseline).
phase: Phase shift (horizontal time shift/ scanner rotation) in same units as `tt`.
offset: Vertical shift (baseline).
peak: Time position of the peak / scanner rotation position, relative to phase and period.
period: Distance between peaks
tss: Total sum of squares.
rss: Residual sum of squares.
R2: Coefficient of determination (pseudo R²).
residual_se: Residual standard error.
df_residual: Degrees of freedom of the residual.
predicted: Fitted values.
residuals: Model residuals.