def ci_umle_boot(X, v, alpha_level):
arr = array_from_data(X, [v])
arr.offset_extremes()
alpha_zero(arr)
fit_model = NonstationaryLogistic()
fit_model.beta['x_0'] = None
fit_model.confidence_boot(arr, alpha_level = alpha_level)
return fit_model.conf['x_0']['pivotal']
covariates = ['x_%d' % i for i in range(1)]
for covariate in covariates:
data_model.beta[covariate] = normal(0, 1.0)
x_node = normal(0, 1.0, N)
def f_x(i_1, i_2):
return abs(x_node[i_1] - x_node[i_2]) < 0.6
net.new_edge_covariate(covariate).from_binary_function_ind(f_x)
net.generate(data_model)
net.offset_extremes()
net.show()
print 'True theta_0: %.2f' % data_model.beta['x_0']
# Initialize the fit model; specify which covariates it should have terms for
fit_model = NonstationaryLogistic()
for covariate in covariates:
fit_model.beta[covariate] = None
# Set up random subnetwork generator, and run fitting experiments
gen = RandomSubnetworks(net, (40, 40))
for rep in range(5):
subnet = gen.sample()
fit_model.fit_brazzale(subnet, 'x_0')
print 'Estimated theta_0: %.2f' % fit_model.beta['x_0']
fit_model.confidence_boot(subnet, n_bootstrap = 10)
cis = fit_model.conf['x_0']
print 'Brazzale CI for theta_0: (%.2f, %.2f)' % cis['brazzale']
print 'Pivotal CI for theta_0: (%.2f, %.2f)' % cis['pivotal']
# Offset extreme substructure only for Nonstationary model
net.offset_extremes()
print 'Fitting nonstationary model'
ns_model = NonstationaryLogistic()
for cov_name in cov_names:
ns_model.beta[cov_name] = None
ns_model.fit(net)
print 'NLL: %.2f' % ns_model.nll(net)
print 'kappa: %.2f' % ns_model.kappa
for cov_name in cov_names:
print '%s: %.2f' % (cov_name, ns_model.beta[cov_name])
print
for rep in range(params['n_samples']):
ns_samples[rep,:,:] = ns_model.generate(net)
ns_model.confidence_boot(net, n_bootstrap = params['n_bootstrap'])
ns_model.confidence_wald(net)
display_cis(ns_model)
# Calculate sample means and variances
s_samples_mean = np.mean(s_samples, axis = 0)
s_samples_sd = np.sqrt(np.var(s_samples, axis = 0))
ns_samples_mean = np.mean(ns_samples, axis = 0)
ns_samples_sd = np.sqrt(np.var(ns_samples, axis = 0))
c_samples_mean = np.mean(c_samples, axis = 0)
c_samples_sd = np.sqrt(np.var(c_samples, axis = 0))
# Finish plotting
plt.figure()
ax = plt.subplot(231)
ax.set_title('Stationary')
data_model.beta[covariate] = normal(0, 1.0)
x_node = normal(0, 1.0, N)
def f_x(i_1, i_2):
return abs(x_node[i_1] - x_node[i_2]) < 0.6
net.new_edge_covariate(covariate).from_binary_function_ind(f_x)
net.generate(data_model)
net.offset_extremes()
net.show()
print 'True theta_0: %.2f' % data_model.beta['x_0']
# Initialize the fit model; specify which covariates it should have terms for
fit_model = NonstationaryLogistic()
for covariate in covariates:
fit_model.beta[covariate] = None
# Set up random subnetwork generator, and run fitting experiments
gen = RandomSubnetworks(net, (40, 40))
for rep in range(5):
subnet = gen.sample()
fit_model.fit_brazzale(subnet, 'x_0')
print 'Estimated theta_0: %.2f' % fit_model.beta['x_0']
fit_model.confidence_boot(subnet, n_bootstrap=10)
cis = fit_model.conf['x_0']
print 'Brazzale CI for theta_0: (%.2f, %.2f)' % cis['brazzale']
print 'Pivotal CI for theta_0: (%.2f, %.2f)' % cis['pivotal']
# Offset extreme substructure only for Nonstationary model
net.offset_extremes()
print 'Fitting nonstationary model'
ns_model = NonstationaryLogistic()
for cov_name in cov_names:
ns_model.beta[cov_name] = None
ns_model.fit(net)
print 'NLL: %.2f' % ns_model.nll(net)
print 'kappa: %.2f' % ns_model.kappa
for cov_name in cov_names:
print '%s: %.2f' % (cov_name, ns_model.beta[cov_name])
print
for rep in range(params['n_samples']):
ns_samples[rep, :, :] = ns_model.generate(net)
ns_model.confidence_boot(net, n_bootstrap=params['n_bootstrap'])
ns_model.confidence_wald(net)
display_cis(ns_model)
# Calculate sample means and variances
s_samples_mean = np.mean(s_samples, axis=0)
s_samples_sd = np.sqrt(np.var(s_samples, axis=0))
ns_samples_mean = np.mean(ns_samples, axis=0)
ns_samples_sd = np.sqrt(np.var(ns_samples, axis=0))
c_samples_mean = np.mean(c_samples, axis=0)
c_samples_sd = np.sqrt(np.var(c_samples, axis=0))
# Finish plotting
plt.figure()
ax = plt.subplot(231)
ax.set_title('Stationary')
s_fit.reset_confidence()
s_fit.confidence_wald(new)
ws_ci_l, ws_ci_u = safe_ci(s_fit, 'x_0', 'wald')
if ws_ci_l < theta < ws_ci_u:
ws_covered += 1
new.offset_extremes()
ns_fit.reset_confidence()
ns_fit.confidence_wald(new)
wn_ci_l, wn_ci_u = safe_ci(ns_fit, 'x_0', 'wald')
if wn_ci_l < theta < wn_ci_u:
wn_covered += 1
ns_fit.confidence_boot(new, n_bootstrap = n_boot,
alpha_level = alpha_level)
bn_ci_l, bn_ci_u = ns_fit.conf['x_0']['pivotal']
if bn_ci_l < theta < bn_ci_u:
bn_covered += 1
A = new.as_dense()
r = A.sum(1)
c = A.sum(0)
c_fit = FixedMargins(s_fit)
new.new_row_covariate('r', np.int)[:] = r
new.new_col_covariate('c', np.int)[:] = c
c_fit.fit = c_fit.base_model.fit_conditional
c_fit.reset_confidence()
c_fit.confidence_wald(new)
