%matplotlib inline
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import scipy.stats
%autosave 30
Autosaving every 30 seconds
Let's have a look how different statistical distributions look like, to have a better idea what to use as prior on our inference bayesian exploration.
All the distributions available in scipy can be found on the docs here: http://docs.scipy.org/doc/scipy/reference/stats.html#module-scipy.stats
Let's start with Discrete distributions
from scipy.stats import bernoulli, poisson, binom
Given a certain probability $p$, the Bernoulli distribution takes value $k=1$, meanwhile it takes $k=0$ in all the other cases $1-p$.
In other words:
$$ f(k;p) = \begin{cases} p & \text{if } k=1 \\\\ 1-p & \text{if } k=0 \end{cases} $$bernoulli.rvs(0.6, size=100)
array([1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0])
a = np.arange(2)
colors = matplotlib.rcParams['axes.color_cycle']
plt.figure(figsize=(12,8))
for i, p in enumerate([0.1, 0.2, 0.6, 0.7]):
ax = plt.subplot(1, 4, i+1)
plt.bar(a, bernoulli.pmf(a, p), label=p, color=colors[i], alpha=0.5)
ax.xaxis.set_ticks(a)
plt.legend(loc=0)
if i == 0:
plt.ylabel("PDF at $k$")
plt.suptitle("Bernoulli probability")
<matplotlib.text.Text at 0x6df5c10>
Another discrete distribution, the Poisson Distribution is defined for all the integer positive number as
$$P(Z=k)=\frac{λ^ke^{−λ}}{k!}, k=0,1,2, \ldots$$k = np.arange(20)
colors = matplotlib.rcParams['axes.color_cycle']
plt.figure(figsize=(12,8))
for i, lambda_ in enumerate([1, 4, 6, 12]):
plt.bar(k, poisson.pmf(k, lambda_), label=lambda_, color=colors[i], alpha=0.4, edgecolor=colors[i], lw=3)
plt.legend()
plt.title("Poisson distribution")
plt.xlabel("$k$")
plt.ylabel("PDF at k")
<matplotlib.text.Text at 0x4642e10>
k = np.arange(15)
plt.figure(figsize=(12,8))
for i, lambda_ in enumerate([1, 2, 4, 6]):
plt.plot(k, poisson.pmf(k, lambda_), '-o', label=lambda_, color=colors[i])
plt.fill_between(k, poisson.pmf(k, lambda_), color=colors[i], alpha=0.5)
plt.legend()
plt.title("Poisson distribution")
plt.ylabel("PDF at $k$")
plt.xlabel("$k$")
<matplotlib.text.Text at 0x5d5ea90>
Last but not least, the binomial distribution which is defined as:
$$f(k;n,p) = Pr(X = k) = {n \choose k} p^k (1-p)^{(n-k)}$$where
$${n \choose k} = \frac{n!}{k!(n-k)!}$$with $k={1, 2, 3, \ldots}$
plt.figure(figsize=(12,6))
k = np.arange(0, 22)
for p, color in zip([0.1, 0.3, 0.6, 0.8], colors):
rv = binom(20, p)
plt.plot(k, rv.pmf(k), lw=2, color=color, label=p)
plt.fill_between(k, rv.pmf(k), color=color, alpha=0.5)
plt.legend()
plt.title("Binomial distribution")
plt.tight_layout()
plt.ylabel("PDF at $k$")
plt.xlabel("$k$")
<matplotlib.text.Text at 0x5ad0710>
They are defined for any value of a positive $x$. A lot of distribution are defined on scipy.stats
, so I will explore only som:
The Alpha distribution is defined as
$$ alpha.pdf(x,a) = \frac{1}{x^2 \Phi(a) \sqrt{2*pi}} * exp(-\frac{1}{2} (\frac{a-1}{x})^2), \,\, with \, x > 0, a > 0 $$x = np.linspace(0.1, 2, 100)
alpha = scipy.stats.alpha
alphas = [0.5, 1, 2, 4]
plt.figure(figsize=(12,6))
for a,c in zip(alphas,colors):
label=r"$\alpha$ = {0:.1f}".format(a)
plt.plot(x, alpha.pdf(x, a), lw=2,
color=c, label=label)
plt.fill_between(x, alpha.pdf(x, a), color=c, alpha = .33)
plt.ylabel("PDF at $x$")
plt.xlabel("$x$")
plt.title("Alpha distribution")
plt.legend()
<matplotlib.legend.Legend at 0x4d61210>
The Beta distribution is defined for a variabile rangin between 0 and 1.
The pdf is defined as:
$$ beta.pdf(x, \alpha, \beta) = \frac{1}{B(\alpha, \beta)}x^{\alpha-1}(1 - x)^{\beta-1}, \; with \; 0≤x≤1, \alpha>0, \beta>0 $$beta = scipy.stats.beta
x = np.linspace(0,1, num=200)
fig = plt.figure(figsize=(12,6))
for a, b, c in zip([0.5, 0.5, 1, 2, 3], [0.5, 1, 3, 2, 5], colors):
plt.plot(x, beta.pdf(x, a, b), lw=2,
c=c, label = r"$\alpha = {0:.1f}, \beta={1:.1f}$".format(a, b))
plt.fill_between(x, beta.pdf(x, a, b), color=c, alpha = .1)
plt.legend(loc=0)
plt.ylabel("PDF at $x$")
plt.xlabel("$x$")
<matplotlib.text.Text at 0x5ad0f50>
The gamma distribution uses the Gamma function (http://en.wikipedia.org/wiki/Gamma_function) and it has two shape parameters.
$$ gamma.pdf(x, \alpha, scale) = \lambda^\alpha * x^{(\alpha-1)} * \frac{exp(-\lambda * x)}{\gamma(\alpha)}, \, with \, x >= 0, \alpha> 0, \lambda > 0 $$The scale parameter is equal = $1.0/\lambda$
gamma = scipy.stats.gamma
plt.figure(figsize=(12, 6))
x = np.linspace(0, 10, num=200)
for a, c in zip([0.5, 1, 2, 3, 10], colors):
plt.plot(x, gamma.pdf(x, a), lw=2,
c=c, label = r"$\alpha = {0:.1f}$".format(a))
plt.fill_between(x, gamma.pdf(x, a), color=c, alpha = .1)
plt.legend(loc=0)
plt.title("Gamma distribution with scale=1")
plt.ylabel("PDF at $x$")
plt.xlabel("$x$")
<matplotlib.text.Text at 0x5ae4810>
The Exponantial probability function is
$$ f_X(x|λ) = λ e^{−λx} , \, x≥0$$Therefore, the random variable X has an exponential distribution with parameter λ, we say X is exponential and write
$$ X∼Exp(λ) $$Given a specific λ, the expected value of an exponential random variable is equal to the inverse of λ, that is:
$$ E[X|λ]= \frac{1}{λ} $$x = np.linspace(0,4, 100)
expo = scipy.stats.expon
lambda_ = [0.5, 1, 2, 4]
plt.figure(figsize=(12,4))
for l,c in zip(lambda_,colors):
plt.plot(x, expo.pdf(x, scale=1./l), lw=2,
color=c, label = "$\lambda = %.1f$"%l)
plt.fill_between(x, expo.pdf(x, scale=1./l), color=c, alpha = .33)
plt.legend()
plt.ylabel("PDF at $x$")
plt.xlabel("$x$")
plt.title("Probability density function of an Exponential random variable;\
differing $\lambda$");