Skip to contents

Fit Two Splines Around a Point of Interest (POI)

Source: R/seek_inflection.R
fit.Rd

This function fits a piecewise linear or parabolic model to melatonin profile data. It separately models the base segment and the ascending segment, ensuring a smooth transition at the point of interest (POI).

Usage

fit(data, poi, fit_type = "linear", threshold = threshold)

Arguments

data

A tibble containing melatonin concentration data with datetime values.

poi

A list containing:

  • x: The x-coordinate (decimal time) of the POI.

  • y: The y-coordinate (melatonin concentration) of the POI.

fit_type

Character. Type of fit for the ascending segment:

  • "linear" (default): Uses a linear model for both base and ascending segments.

  • "parabolic": Uses a parabolic fit for the ascending segment.

threshold

Numeric. The threshold for constraining the base segment.

Value

A list containing:

  • residual: The sum of squared residuals from both fits.

  • base_params: The estimated parameters for the base fit.

  • ascending_params: The estimated parameters for the ascending fit.

Details

  • The base segment is always fit using a linear model.

  • The ascending segment is fit using a linear or parabolic model depending on fit_type.

  • If the ascending segment has at least 3 data points, it first fits a linear model and then refines it with a parabolic fit.

  • This function is used in inflection point detection.

See also

fit_profile, objective_function, seek_inflection

Examples

data <- tibble::tibble(
  datetime = as.POSIXct(c("2024-06-01 22:00", "2024-06-01 22:15", "2024-06-01 22:30")),
  melatonin = c(0.5, 0.8, 1.2),
  base = c(1, 1, 0)  # Base segment indicators
)
poi <- list(x = posixct_to_decimal(data$datetime[2], data$datetime), y = data$melatonin[2])
fit(data, poi, fit_type = "linear", threshold = 2.3)
#> $residual
#> [1] 0.03125
#> 
#> $base_params
#> [1] 0.2
#> 
#> $ascending_params
#> [1] 1.6
#>