Sorting Data

Some types of characterization data, such as FTIR and XPS, are typically expressed with their x-values in decending order. Rather than having to ensure users rearrange their data into ascending order before using pybaselines, the Baseline object handles this internally and returns values in the same order as the input x-values. This is especially important if other parameters from the selected baseline method are desired such as weights, since those are likewise returned in the same order as the input x-values.

import matplotlib.pyplot as plt
import numpy as np

from pybaselines import Baseline
from pybaselines.utils import gaussian


x = np.linspace(500, 4000, 1000)
signal = (
    + gaussian(x, 8, 650, 16)
    + gaussian(x, 9, 1100, 50)
    + gaussian(x, 8, 1350, 20)
    + gaussian(x, 11, 2800, 20)
    + gaussian(x, 8, 2900, 20)
    + gaussian(x, 5, 3400, 120)
)
baseline = 0.08 + 0.00004 * x + gaussian(x, 3, 3500, 1000)
noise = np.random.default_rng(0).normal(0, 0.1, len(x))
y = signal + baseline + noise

# reverse both x and y
x_reversed = x[::-1]
y_reversed = y[::-1]
lam = 1e6

fit, params = Baseline(x).iarpls(y, lam=lam)
fit_reversed, params_reversed = Baseline(x_reversed).iarpls(y_reversed, lam=lam)

_, (ax_1, ax_2) = plt.subplots(nrows=2, sharex=True)
ax_1.plot(x, y)
ax_1.plot(x, fit)
ax_2.plot(x_reversed, y_reversed)
ax_2.plot(x_reversed, fit_reversed)
ax_2.invert_xaxis()
ax_2.set_xlabel(r'Wavenumber (cm$^{-1}$)')
ax_1.set_ylabel('Absorbance (%)')
ax_2.set_ylabel('Absorbance (%)')

ax_1.set_title('Sorted Data')
ax_2.set_title('Reversed Data')

print(
    'Outputs same for sorted and unsorted baselines: ',
    np.allclose(fit, fit_reversed[::-1], rtol=1e-14, atol=1e-14)
)

Outputs same for sorted and unsorted baselines:  True

Likewise, relevant output parameters such as weights are also sorted to match the input x-values.

_, (ax_1, ax_2) = plt.subplots(nrows=2, sharex=True)
ax_1.plot(x, y)
ax_1.plot(x, fit)
ax_2.plot(x, params['weights'], 'ro', label='sorted')
ax_2.plot(x_reversed, params_reversed['weights'], 'b.', label='reversed')
ax_2.invert_xaxis()
ax_2.legend()

ax_2.set_xlabel(r'Wavenumber (cm$^{-1}$)')
ax_1.set_ylabel('Absorbance (%)')
ax_2.set_ylabel('Weights')

print(
    'Weights same for sorted and unsorted baselines: ',
    np.allclose(params['weights'], params_reversed['weights'][::-1], rtol=1e-14, atol=1e-14)
)

plt.show()

Weights same for sorted and unsorted baselines:  True