Skip to main content Link Menu Expand (external link) Document Search Copy Copied

Scale

Scale is a preprocessing technique in spectroscopy that scales the spectra. The following algorithms are available:

Point scaler

Point scaler is a preprocessing technique in spectroscopy that scales each spectrum by the absorbance/intensity at a given index or wavenumber. The default index is 0, which means that the first point of each spectrum is used for scaling.

Arguments:

Argument Description Type Default
point The index or wavenumber of the spectrum to use for scaling. If not wavenumbers it will use the index. int 0
wavenumber The wavenumbers of the spectra. Optional. np.ndarray/list None

The wavenumbers vector must be sorted in ascending order.

Usage examples: 

from chemotools.scale import PointScaler

point_scaler = PointScaler(point=310)
spectra_scaled = point_scaler.fit_transform(spectra)

Plotting example:

MinMax scaler

MinMaxScaler is a preprocessing technique in spectroscopy subtracts the minimum value of the spectrum and divides it by the difference between the maximum and the minimum value of the spectrum. If the parameter use_min is set to False, the spectrum is just divided by the maximum value of the spectrum.

Arguments:

Argument Description Type Default
use_min If True, the spectrum is subtracted by its minimum value and divided by the difference between the maximum and the minimum. If False, the spectrum is scaled by its maximum value. bool True

Usage examples: 

from chemotools.scale import MinMaxScaler

minmax = MinMaxScaler()
spectra_norm = minmax.fit_transform(spectra)

Plotting example:

L-norm scaler

L-normalization is a preprocessing technique in spectroscopy that scales each spectrum by its L-norm.

Arguments:

Argument Description Type Default
l_norm The L-norm to use. int 2

Usage examples: 

from chemotools.scale import NormScaler

lnorm = NormScaler(l_norm=2)
spectra_norm = lnorm.fit_transform(spectra)

Plotting example: