import numpy as np
import bayes_factor
if __name__=='__main__':
iterations = 1e4
N = 15
m = 108
errors = np.array([43, 59, 51, 38, 39, 53, 47, 50, 50, 59, 59, 45, 36, 46, 53])
# errors = np.linspace(0,60,N).astype(np.int)
print "errors =", errors
B_10, p_errors_given_mu_sigma, p_errors_given_sigma, sigma0, mu1, sigma1 = bayes_factor.bayes_factor(errors, m, iterations)
sigma1_upper = np.sqrt(mu1 * (1.0 - mu1))
p_mu_sigma_given_errors = p_errors_given_mu_sigma * 1.0/sigma1_upper * 2.0
integral_mc1 = p_errors_given_mu_sigma.mean()
p_mu_sigma_given_errors /= integral_mc1
p_sigma_given_errors = p_errors_given_sigma * 2.0
integral_mc0 = p_errors_given_sigma.mean()
p_sigma_given_errors /= integral_mc0
import matplotlib.mlab as mlab
import matplotlib.pyplot as plt
size = 500
sigma_i = np.linspace(0.0, 0.5, size)
print "m_range =", m_range
print "N_range =", N_range
print "mu =", mu_true
print "sigma =", sigma_true
print "M =", M
BF = np.zeros((M, len(m_range), len(N_range)))
for k, N in enumerate(N_range):
for j, m in enumerate(m_range):
for i in range(M):
epsilon = np.random.beta(a=a, b=b, size=N)
errors = np.random.binomial(n=m, p=epsilon)
print i,') N =', N, ', m =', m
print "\t errors:", errors
B_10, p1, p0, a0, a1, b1 = bayes_factor.bayes_factor(errors, m, iterations)
BF[i,j,k] = B_10
print "\t p(errors|H0) =", p0.mean()
print "\t p(errors|H1) =", p1.mean()
print "\t B_10 =", B_10
print "\t B_01 =", 1.0/B_10
BF_threshold = [1, 3, 20, 150]
import matplotlib.pyplot as plt
if mu_true==0.5:
description = "p(Type I error)"
ylabel = description
else:
description = "p(Type II error)"
