Plot mixture of univariate distributions

plot_mixture.py

from scipy.stats import norm

import matplotlib.pyplot as plt
import numpy as np


plt.style.use("ggplot")


# Simulate from a Univariate Gaussian Mixture Model.
rs = np.random.RandomState(13)

n_components = 3
n_draws = 1000
weights = np.array([.5, .2, .3])
means = np.array([25, 10.8, 32.1])
stds = np.array([1.75, 1, 2.2])

labels = rs.choice(n_components, size=n_draws, replace=True, p=weights)
data = np.array([
    rs.normal(loc=means[k], scale=stds[k])
    for k in labels
])


# Plot each distribution against the histogram.
fig, ax = plt.subplots()
ax.hist(data, bins=100, density=True, edgecolor="k")

xpts = np.linspace(data.min(), data.max(), 250)
for k in range(n_components):
    densities = norm.pdf(xpts, loc=means[k], scale=stds[k])
    ax.plot(xpts, densities, label=f"k={k + 1}")

ax.legend()

fig.savefig("univariate-mixture.png")
plt.close(fig)

univariate-mixture.png