General purpose preprocessing of spectra for predictions

irp_preprocess is a function that provides a general-purpose preprocessing workflow for spectra. The workflow comprises interpolation, clipping, baseline correction, smoothing (including optionally derivatization), normalization, binning, and scaling. All these steps are optionally, but occur on a fixed order that cannot be changed.

Usage

irp_preprocess(
  x,
  do_interpolate = TRUE,
  interpolate_start = NULL,
  interpolate_dw = 1,
  do_clip = TRUE,
  clip_range,
  do_interpolate_region = FALSE,
  interpolate_region_range,
  do_bc = TRUE,
  bc_method = "rubberband",
  bc_degree = 2,
  bc_cutoff = 0,
  bc_do_impute = FALSE,
  do_smooth = TRUE,
  smooth_method = "sg",
  smooth_p = 3,
  smooth_n = smooth_p + 3 - smooth_p%%2,
  smooth_m = 1,
  smooth_ts = 0,
  smooth_k = 111,
  do_normalise = TRUE,
  normalise_method = "area",
  do_bin = TRUE,
  bin_width = 10,
  bin_new_x_type = "start",
  do_scale = TRUE,
  scale_center = TRUE,
  scale_scale = TRUE
)

Arguments

x: An object of class ir (it is assumed that x is not yet preprocessed).
do_interpolate: A logical value indicating if spectra should be interpolated using ir::ir_interpolate().
interpolate_start: See ir_interpolate (parameter start).
interpolate_dw: See ir_interpolate (parameter dw).
do_clip: A logical value indicating if spectra should be clipped using ir::ir_clip().
clip_range: See ir_clip (parameter range).
do_interpolate_region: A logical value indicating if spectra should be linearly interpolated in selected regions using ir::ir_interpolate_region().
interpolate_region_range: See ir_interpolate_region (parameter range).
do_bc: A logical value indicating if spectra should be baseline corrected using ir::ir_bc().
bc_method: See ir_bc (parameter method).
bc_degree: See ir_bc (parameter degree).
bc_cutoff: A numeric value representing the wavenumber units to remove at the start and end of each spectrum in x during baseline correction. This may be done to remove artifacts due to baseline correction.
bc_do_impute: See ir_bc (parameter do_impute).
do_smooth: A logical value indicating if spectra should be smoothed using ir::ir_smooth().
smooth_method: See ir_smooth (parameter method).
smooth_p: See ir_smooth (parameter p).
smooth_n: See ir_smooth (parameter n).
smooth_m: See ir_smooth (parameter m).
smooth_ts: See ir_smooth (parameter ts).
smooth_k: See ir_smooth (parameter k).
do_normalise: A logical value indicating if spectra should be normalized using ir::ir_normalize().
normalise_method: See ir_normalize (parameter method).
do_bin: A logical value indicating if spectra should be binned using ir::ir_bin().
bin_width: See ir_bin (parameter width).
bin_new_x_type: See ir_bin (parameter new_x_type).
do_scale: A logical value indicating if spectral variables should be scaled using base::scale().
scale_center: See scale (parameter center). To scale each spectral variable independently, provide a vector with length equal to the number of spectral variables returned after preprocessing.
scale_scale: See scale (parameter scale). To scale each spectral variable independently, provide a vector with length equal to the number of spectral variables returned after preprocessing.

Value

A data frame with spectra in rows and a column for each spectral variable after preprocessing.

Examples

# get sample data
x <- ir::ir_sample_data[1, ]

# example preprocessing
res <-
  irpeat::irp_preprocess(
    x,
    do_interpolate = TRUE,
    interpolate_start = NULL,
    interpolate_dw = 1,
    do_bc = TRUE,
    do_clip = FALSE,
    do_interpolate_region = FALSE,
    bc_method = "rubberband",
    bc_cutoff = 10,
    bc_do_impute = FALSE,
    do_smooth = FALSE,
    do_normalise = TRUE,
    normalise_method = "area",
    do_bin = TRUE,
    bin_width = 10,
    bin_new_x_type = "start",
    do_scale = TRUE,
    scale_center = TRUE,
    scale_scale = TRUE
  )