Classification masks

The baseline algorithms in the classification module estimate the baseline by classifying each point as belonging to either the baseline or the peaks. When first using a function, the correct parameters may not be known. To make the effects of input parameters on the classification process more easily understood, all functions in the classification module provide a 'mask' item in the output parameter dictionary. The mask parameter is a boolean numpy array that is True for any point classified as belonging to the baseline and False otherwise.

import matplotlib.pyplot as plt
import numpy as np

from pybaselines import Baseline
from pybaselines.utils import gaussian


half_window = 50
x = np.linspace(0, 1000, 1000)
signal = (
    gaussian(x, 9, 100, 12)
    + gaussian(x, 6, 180, 5)
    + gaussian(x, 8, 350, 11)
    + gaussian(x, 15, 400, 18)
    + gaussian(x, 6, 550, 6)
    + gaussian(x, 13, 700, 8)
    + gaussian(x, 9, 800, 9)
    + gaussian(x, 9, 880, 7)
)
baseline = gaussian(x, -6, 700, 500)
noise = np.random.default_rng(0).normal(0, 0.1, x.size)
y = signal + baseline + noise

baseline_fitter = Baseline(x_data=x)

When first fitting a new dataset, it may be difficult to estimate the correct parameters. For this example, the main parameter for the baseline function is the half_window used for the rolling standard deviation calculation. Try a low and high value to see the difference.

half_window_1 = 15
half_window_2 = 45
fit_1, params_1 = baseline_fitter.std_distribution(y, half_window_1, smooth_half_window=10)
fit_2, params_2 = baseline_fitter.std_distribution(y, half_window_2, smooth_half_window=10)

plt.plot(x, y)
plt.plot(x, fit_1, label=f'half_window={half_window_1}')
plt.plot(x, fit_2, '--', label=f'half_window={half_window_2}')
plt.legend()

The two baselines are similar in most regions except for the two small peaks. To investigate why such different results were obtained, the mask item in the output parameter dictionary can be used.

mask_1 = params_1['mask']
mask_2 = params_2['mask']

_, (ax, ax2) = plt.subplots(2, sharex=True, gridspec_kw={'hspace': 0})
ax.plot(x, y)
patch_1 = ax.plot(x[mask_1], y[mask_1], 'o')[0]
ax2.plot(x, y)
patch_2 = ax2.plot(x[mask_2], y[mask_2], 'ms')[0]
ax.legend((patch_1, patch_2), (f'half_window={half_window_1}', f'half_window={half_window_2}'))

plt.show()

After comparing the two masks, it is clear that the higher half_window value mis-identified the small peaks as belonging to the baseline. Thus, the smaller half_window is the better parameter.