liner = pm.Deterministic('liner', tt.exp(u + beta + (beta1[companyABC] * elec_year + \
beta2[companyABC] * elec_Pca_char1 + beta3[
companyABC] * elec_Pca_char2 + \
beta4[companyABC] * elec_year * elec_year)))
phic = pm.Uniform('phic', lower=0, upper=1, testval=.3)
zck = pm.Bernoulli('zck', p=phic, shape=companyABC.shape)
# zij_ = pm.theanof.tt_rng().uniform(size=xij.shape)
# zij = pm.Deterministic('zij', tt.lt(zij_, phii[sbjid]))
line = tt.constant(np.ones((len(companyABC))) * .5)
beta_mu = pm.Deterministic(
'beta_mu', tt.squeeze(tt.switch(tt.eq(zck, 0), liner, line)))
Observed = pm.Weibull("Observed",
alpha=alpha,
beta=beta_mu,
observed=elec_faults) # 观测值
# start = pm.find_MAP()
step = pm.Metropolis([zck])
# step1 = pm.Slice([am0, am1])
trace = pm.sample(4000, step=[step], init='advi+adapt_diag', tune=1000)
# with model1:
# s = shared(pm.floatX(1))
# inference = pm.ADVI(cost_part_grad_scale=s)
# # ADVI has nearly converged
# inference.fit(n=20000)
# # It is time to set `s` to zero
# s.set_value(0)
# approx = inference.fit(n=10000)
# 这里将威布尔参数的alpha 与 beta交换
with pm.Model() as unpooled_model:
sigma = pm.HalfCauchy('sigma', 10, testval=1.)
# BoundedNormal = pm.Bound(pm.Normal, lower=3)
# switchpoint = BoundedNormal('switchpoint', 5, 5)
# switchpoint = pm.HalfNormal('switchpoint', 2)
early_rate = pm.Normal('early_rate', 0, 100)
late_rate = pm.Normal('late_rate', 0, 100)
beta1 = pm.math.switch(x_shared <= 5, early_rate, late_rate)
beta = pm.Normal('beta', 0, 100)
u = pm.Normal('u', 0, 0.0001)
mu = pm.Deterministic('mu', tt.exp(beta + beta1 * x_shared + u))
# Observed_pred = pm.Weibull("Observed_pred", alpha=mu, beta=sigma, shape=elec_faults.shape) # 观测值
Observed = pm.Weibull("Observed", alpha=sigma, beta=mu,
observed=y_shared) # 观测值
start = pm.find_MAP()
# step = pm.Metropolis([switchpoint])
trace2 = pm.sample(3000, start=start)
chain2 = trace2[1000:]
varnames2 = ['beta', 'early_rate', 'late_rate', 'sigma', 'u']
pm.traceplot(chain2)
plt.show()
pm.energyplot(trace2)
plt.show()
# # 画出自相关曲线
# pm.autocorrplot(chain2, varnames2)
# plt.show()
print(pm.dic(trace2, unpooled_model))
# 调好的基于weibull的模型，但是后面分析有点问题，拟合效果不明显
with pm.Model() as unpooled_model:
# define priors
sigma = pm.Gamma('sigma', 1, 1.6)
# mu = pm.Gamma('mu', 2, 5)
beta = pm.Gamma('beta', 20, 2000, shape=companiesABC)
beta1 = pm.Gamma('beta1', 20, 2000, shape=companiesABC)
beta2 = pm.Gamma('beta2', 4, 80000)
mu = pm.Deterministic(
'mu',
tt.exp(beta[companyABC] + beta1[companyABC] * elec_year +
beta2 * elec_tem))
Observed_pred = pm.Weibull("Observed_pred",
alpha=mu,
beta=sigma,
shape=elec_faults.shape) # 观测值
Observed = pm.Weibull("Observed",
alpha=sigma,
beta=mu,
observed=elec_faults) # 观测值beta为尺度参数
start = pm.find_MAP()
# step = pm.Metropolis()
trace2 = pm.sample(2000, start=start)
chain2 = trace2
# varnames1 = ['beta', 'beta1', 'sigma', 'mu']
pm.traceplot(chain2)
plt.show()
com_pred = chain2.get_values('Observed_pred')[:].ravel()
beta = pm.Gamma('beta', 2, 200, shape=companiesABC)
beta1 = pm.Gamma('beta1', 2, 200, shape=companiesABC)
beta2 = pm.Gamma('beta2', 4, 80000)
# beta = pm.Normal('beta', 5, 50, shape=companiesABC)
# beta1 = pm.Normal('beta1', 0, 10, shape=companiesABC)
# beta2 = pm.Normal('beta2', 0, 20)
# u = pm.Gamma('u', 2, 200)
# mu = tt.exp(beta[companyABC] + beta1[companyABC]*elec_year + beta2*elec_tem)
mu = pm.Deterministic(
'mu',
tt.exp(beta[companyABC] + beta1[companyABC] * elec_year +
beta2 * elec_tem))
# Observed_pred = pm.Weibull("Observed_pred", alpha=mu, beta=sigma, shape=elec_faults.shape) # 观测值
Observed = pm.Weibull("Observed",
alpha=mu,
beta=sigma,
observed=elec_faults) # 观测值
# start = pm.find_MAP()
# step = pm.Metropolis([Observed])
trace2 = pm.sample(1000)
chain2 = trace2
varnames1 = ['sigma', 'mu']
pm.traceplot(chain2)
plt.show()
print(pm.dic(trace2, unpooled_model))
print(pm.df_summary(trace2, varnames1))
# com_pred = chain2.get_values('Observed_pred')[:].ravel()
# plt.hist(com_pred, range=[0, 2], bins=130, histtype='stepfilled')
# plt.axvline(elec_faults.mean(), color='r', ls='--', label='True mean')
# theta = pm.InverseGamma('theta', 0.01, 0.01)
# alpha = pm.Normal('alpha', 0, theta)
# 建立与时间相关的函数
# define likelihood
out_pai = pm.Deterministic(
'out_pai',
np.exp(mu + beta * dag_time + gamma1 * z1 + gamma2 * z2 + gamma3 * z3 +
gamma4 * z4) /
(1 + np.exp(mu + beta * dag_time + gamma1 * z1 + gamma2 * z2 +
gamma3 * z3 + gamma4 * z4)))
# out_pai = pm.math.invlogit(mu + alpha + beta * dag_time + gamma1 * z1 + gamma2 * z2 + gamma3 * z3 + gamma4 * z4)
# out_pai = pm.Deterministic('out_pai', mu + beta * dag_time + gamma1 * z1 + gamma2 * z2 + gamma3 * z3 + gamma4 * z4)
Observed = pm.Weibull("Observed", 1, out_pai, observed=dag_faultpro) # 观测值
start = pm.find_MAP()
# step = pm.Metropolis()
trace = pm.sample(10000, start=start)
chain = trace
# logistic(chain, locals())
varnames = ['gamma1', 'mu', 'beta', 'out_pai']
varnames1 = ['out_pai']
pm.traceplot(chain, varnames)
plt.show()
# 画出自相关曲线
pm.autocorrplot(chain)
plt.show()
beta = pm.Normal('beta', 0, 100, shape=companiesABC, testval=-3.)
# beta1 = pm.Normal('beta1', 0, 10, shape=companiesABC, testval=.3)
# beta2 = pm.Normal('beta2', 0, 100, testval=0.01)
# beta3 = pm.Normal('beta3', 0, 100)
theta = pm.Normal('theta', 0, 100, shape=companiesABC)
theta1 = pm.Normal('theta1', 0, 20, shape=companiesABC)
beta1 = theta[companyABC] + theta1[companyABC] * x_shared1
# mu = tt.exp(beta[companyABC] + beta1[companyABC]*elec_year + beta2*elec_tem)
beta_mu = pm.Deterministic(
'beta_mu', tt.exp(beta[companyABC] + beta1[companyABC] * x_shared))
# Observed_pred = pm.Weibull("Observed_pred", alpha=mu, beta=sigma, shape=elec_faults.shape) # 观测值
Observed = pm.Weibull("Observed",
alpha=alpha,
beta=beta_mu,
observed=y_shared) # 观测值
start = pm.find_MAP()
# step = pm.Slice([beta1, u])
trace2 = pm.sample(2000, start=start)
chain2 = trace2[1000:]
varnames1 = ['alpha', 'beta_mu']
# varnames2 = ['beta', 'beta1', 'beta2', 'alpha', 'beta3']
# pm.plot_posterior(chain2, varnames2, ref_val=0)
pm.traceplot(chain2)
plt.show()
pm.energyplot(trace2)
plt.show()
# elec_year1 = np.delete(elec_year, -1)
theta1 = pm.Deterministic('theta1', a0 + (σ_a * Δ_a).cumsum())
# theta = pm.Deterministic('theta', tt.exp(Bx_.dot(theta1) + δ + beta*elec_tem1[0:84]+ beta1*elec_RH1[0:84]))
theta = pm.Deterministic('theta', tt.exp(Bx_.dot(theta1) + δ))
theta1B = pm.Deterministic('theta1B', a0B + (σ_aB * Δ_aB).cumsum())
thetaB = pm.Deterministic('thetaB', tt.exp(Bx_B.dot(theta1B) + δB))
theta1C = pm.Deterministic('theta1C', a0C + (σ_aC * Δ_aC).cumsum())
thetaC = pm.Deterministic('thetaC', tt.exp(Bx_C.dot(theta1C) + δC))
# Observed = pm.Poisson('Observed', mu =theta, observed=ys_faultsA) # 观测值
# ObservedB = pm.Poisson('ObservedB', mu =thetaB, observed=ys_faultsB) # 观测值
# ObservedC = pm.Poisson('ObservedC', mu =thetaC, observed=ys_faultsC) # 观测值
Observed = pm.Weibull('Observed',
alpha=alpha,
beta=theta,
observed=ys_faultsA) # 观测值
ObservedB = pm.Weibull('ObservedB',
alpha=alpha,
beta=thetaB,
observed=ys_faultsB) # 观测值
ObservedC = pm.Weibull('ObservedC',
alpha=alpha,
beta=thetaC,
observed=ys_faultsC) # 观测值
# step1 = pm.Slice([Δ_a])
start = pm.find_MAP()
trace_1 = pm.sample(1000, start=start, njobs=1)
# ================================================================================
# Return a B-spline basis element B(x | t[0], ..., t[k+1])