Module maven_iuvs.statistics
Expand source code
import numpy as np
from sklearn import linear_model
def multiple_linear_regression(templates, spectrum, spectrum_error):
"""
Fits an array of templates to a spectrum using multiple linear regression (MLR).
Parameters
----------
templates : list, arr
Templates to fit to the spectrum in DN.
spectrum : list, arr
An observed spectrum in DN.
spectrum_error : list, arr
The uncertainty on the spectrum values in DN.
Returns
-------
coeff : float, arr
The fit coefficients for the templates
const : float
The constant term from the fitting.
"""
# ensure input templates are an array with two dimensions
X = np.array(templates)
if X.ndim == 1:
np.expand_dims(X, axis=0)
# transpose templates
X = X.T
# ensure spectrum and error are numpy arrays
Y = np.array(spectrum)
Yerr = np.array(spectrum_error)
# convert uncertainty to sample weight
Yw = (1/Yerr)**2
# make a linear regression model and fit the spectra with templates
fit = linear_model.LinearRegression().fit(X, Y, sample_weight=Yw)
# extract the coefficients and constant term
coeff = fit.coef_
const = fit.intercept_
# return the coefficients and constant term
return coeff, const
def integrate_intensity(template_wavelength, template_spectrum, calibration_curve, coefficient):
"""
Takes a spectrum template and MLR coefficient and calculates an integrated intensity.
Parameters
----------
template_wavelength : array
Spectrum wavelengths.
template_spectrum : array
Spectrum in DN.
calibration_curve : array
Conversion from DN to physical units. Assumes format is [DN/unit].
coefficient : float
MLR fit coefficient.
Returns
-------
integrated_intensity : float
Integrated template intensity in physical units.
"""
# determine spectral bin spacing
dwavelength = np.diff(template_wavelength)[0]
# calibrate the template
calibrated_template_spectrum = coefficient * template_spectrum / calibration_curve
# integrate the template
integrated_intensity = np.trapz(calibrated_template_spectrum, dx=dwavelength)
# return the integrated intensity
return integrated_intensityFunctions
def integrate_intensity(template_wavelength, template_spectrum, calibration_curve, coefficient)-
Takes a spectrum template and MLR coefficient and calculates an integrated intensity.
Parameters
template_wavelength:array- Spectrum wavelengths.
template_spectrum:array- Spectrum in DN.
calibration_curve:array- Conversion from DN to physical units. Assumes format is [DN/unit].
coefficient:float- MLR fit coefficient.
Returns
integrated_intensity:float- Integrated template intensity in physical units.
Expand source code
def integrate_intensity(template_wavelength, template_spectrum, calibration_curve, coefficient): """ Takes a spectrum template and MLR coefficient and calculates an integrated intensity. Parameters ---------- template_wavelength : array Spectrum wavelengths. template_spectrum : array Spectrum in DN. calibration_curve : array Conversion from DN to physical units. Assumes format is [DN/unit]. coefficient : float MLR fit coefficient. Returns ------- integrated_intensity : float Integrated template intensity in physical units. """ # determine spectral bin spacing dwavelength = np.diff(template_wavelength)[0] # calibrate the template calibrated_template_spectrum = coefficient * template_spectrum / calibration_curve # integrate the template integrated_intensity = np.trapz(calibrated_template_spectrum, dx=dwavelength) # return the integrated intensity return integrated_intensity def multiple_linear_regression(templates, spectrum, spectrum_error)-
Fits an array of templates to a spectrum using multiple linear regression (MLR).
Parameters
templates:list, arr- Templates to fit to the spectrum in DN.
spectrum:list, arr- An observed spectrum in DN.
spectrum_error:list, arr- The uncertainty on the spectrum values in DN.
Returns
coeff:float, arr- The fit coefficients for the templates
const:float- The constant term from the fitting.
Expand source code
def multiple_linear_regression(templates, spectrum, spectrum_error): """ Fits an array of templates to a spectrum using multiple linear regression (MLR). Parameters ---------- templates : list, arr Templates to fit to the spectrum in DN. spectrum : list, arr An observed spectrum in DN. spectrum_error : list, arr The uncertainty on the spectrum values in DN. Returns ------- coeff : float, arr The fit coefficients for the templates const : float The constant term from the fitting. """ # ensure input templates are an array with two dimensions X = np.array(templates) if X.ndim == 1: np.expand_dims(X, axis=0) # transpose templates X = X.T # ensure spectrum and error are numpy arrays Y = np.array(spectrum) Yerr = np.array(spectrum_error) # convert uncertainty to sample weight Yw = (1/Yerr)**2 # make a linear regression model and fit the spectra with templates fit = linear_model.LinearRegression().fit(X, Y, sample_weight=Yw) # extract the coefficients and constant term coeff = fit.coef_ const = fit.intercept_ # return the coefficients and constant term return coeff, const