Created
November 3, 2013 18:11
-
-
Save aniversarioperu/7293034 to your computer and use it in GitHub Desktop.
Análisis estadístico bayesiano del número de delitos por año para Perú. Se obtiene el número esperado de delitos y el año más probable en que se produjo un aumento en la cantidad de delitos. Usando teoría y cógidos de "Bayesian methods for hackers": http://bit.ly/19tyAUX
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
# -*- coding: utf-8 -*- | |
import prettyplotlib as ppl | |
from prettyplotlib import plt | |
import sys | |
import pymc as pm | |
import numpy as np | |
datos = np.loadtxt("datos.txt") | |
alpha = 1.0/datos.mean() | |
print alpha | |
print "alpha %f" % alpha | |
print "datos.mean %f" % datos.mean() | |
n_datos = len(datos) | |
lambda_1 = pm.Exponential("lambda_1", alpha) | |
lambda_2 = pm.Exponential("lambda_2", alpha) | |
print lambda_1.random() | |
print lambda_2.random() | |
tau = pm.DiscreteUniform("tau", lower=0, upper=n_datos) | |
print tau.random() | |
@pm.deterministic | |
def lambda_(tau=tau, lambda_1=lambda_1, lambda_2=lambda_2): | |
out = np.zeros(n_datos) | |
out[:tau] = lambda_1 | |
out[tau:] = lambda_2 | |
return out | |
observation = pm.Poisson("obs", lambda_, value=datos, observed=True) | |
model = pm.Model([observation, lambda_1, lambda_2, tau]) | |
mcmc = pm.MCMC(model) | |
mcmc.sample(50000, 10000, 1) | |
lambda_1_samples = mcmc.trace('lambda_1')[:] | |
lambda_2_samples = mcmc.trace('lambda_2')[:] | |
tau_samples = mcmc.trace('tau')[:] | |
fig, (ax1, ax2, ax3) = plt.subplots(nrows=3, ncols=1) | |
plt.rc('font', **{'family': 'DejaVu Sans'}) | |
plt.subplot(311) | |
plt.title(u'''Distribución posterior de las variables | |
$\lambda_1,\;\lambda_2,\;tau$''') | |
plt.hist(lambda_1_samples, histtype="stepfilled", bins=30, alpha=0.85, | |
normed=True) | |
plt.xlim([150000,250000]) | |
plt.xlabel("valor de $\lambda_1$") | |
plt.subplot(312) | |
#ax.set_autoscaley_on(False) | |
plt.hist(lambda_2_samples, histtype="stepfilled", bins=30, alpha=0.85, | |
normed=True) | |
plt.xlim([150000,250000]) | |
plt.xlabel("valor de $\lambda_2$") | |
plt.tick_params(axis="both", which="mayor", labelsize=4) | |
plt.subplot(313) | |
w = 1.0/tau_samples.shape[0]*np.ones_like(tau_samples) | |
plt.hist(tau_samples, bins=n_datos, alpha=1, weights=w, rwidth=2.0) | |
plt.xticks(np.arange(n_datos)) | |
plt.ylim([0, 1.5]) | |
plt.xlim([0, 8]) | |
plt.xlabel("valor de $tau$") | |
fig.set_size_inches(7,6) | |
fig.tight_layout() | |
fig.savefig("plot1.png") | |
fig, ax = plt.subplots(nrows=1, ncols=1) | |
N = tau_samples.shape[0] | |
expected_texts_per_day = np.zeros(n_datos) | |
for day in range(0, n_datos): | |
ix = day < tau_samples | |
expected_texts_per_day[day] = (lambda_1_samples[ix].sum() | |
+ lambda_2_samples[~ix].sum()) / N | |
anhos = ["2005","2006","2007","2008","2009","2010","2011","2012"] | |
plt.plot(range(n_datos), expected_texts_per_day, lw=4, color="#E24A33", | |
label="expected number of text-messages received") | |
plt.xlim(0, n_datos) | |
plt.xticks(np.arange(n_datos) + 0.4, anhos) | |
plt.xlabel(u'Años') | |
plt.ylabel(u'Número esperado de delitos') | |
plt.title(u'''Cambio en el número esperado de delitos por año''') | |
plt.ylim(0, 300000) | |
plt.bar(np.arange(len(datos)), datos, color="#348ABD", alpha=0.65) | |
#plt.legend(loc="upper left") | |
fig.savefig("plot2.png") |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment