Vectorized Inputs

All data and priors will allow for vectorized assuming the shapes work for broadcasting.

The plotting also supports arrays of results

import numpy as np

from conjugate.distributions import Beta
from conjugate.models import binomial_beta

import matplotlib.pyplot as plt

# Data
N = 10
x = 4

# Analytics
prior = Beta(alpha=1, beta=np.array([1, 10]))
posterior = binomial_beta(n=N, x=x, prior=prior)

# Figure
colors = ["blue", "red"]
ax = prior.plot_pdf(label=lambda i: f"prior {i}", color=colors, linestyle="--")
posterior.plot_pdf(ax=ax, label=lambda i: f"posterior {i}", color=colors)
ax.axvline(x=x / N, ymax=0.05, color="black", linestyle="--", label="MLE")
ax.legend()
plt.show()

Comments