top of page

From: Bayesian Models for Astrophysical Data, Cambridge Univ. Press

(c) 2017,  Joseph M. Hilbe, Rafael S. de Souza and Emille E. O. Ishida

you are kindly asked to include the complete citation if you used this material in a publication

â€‹

â€‹

Code 7.5 Bayesian zero-inflated negative binomial model in Python using Stan

=================================================================

import numpy as np
import pystan
import statsmodels.api as sm

from rpy2.robjects import r, FloatVector
from scipy.stats import uniform, bernoulli

def gen_zinegbinom(N, mu1, mu2, alpha):
"""Zero inflated negative binomial distribution."""

r('require(VGAM)')

# get R functions
zinbinomR = r['rzinegbin'
res = zinbinomR(n=N, munb=FloatVector(mu1), size=1.0/alpha,
pstr0=FloatVector(mu2))

â€‹

return np.array([int(item) for item in res])

# Data
np.random.seed(141)                                              # set seed to replicate example
nobs= 7500                                                             # number of obs in model

â€‹

x1 = uniform.rvs(size=nobs)
x2 = bernoulli.rvs(0.6, size=nobs)

xb = 1.0 + 2.0 * x1 + 1.5 * x2                                # linear predictor
xc = 2.0 - 5.0 * x1 + 3.0 * x2

â€‹

exb = np.exp(xb)
exc = 1 / (1 + np.exp(-xc))
alpha = 2

â€‹

zinby = gen_zinegbinom(nobs, exb, exc, alpha)

â€‹

X = np.column_stack((x1,x2))

â€‹

mydata = {}                                                              # build data dictionary
mydata['Y'] = zinby                                                  # response variable
mydata['N'] = nobs                                                   # sample size
mydata['Xb'] = X                                                      # predictors
mydata['Xc'] = X
mydata['Kb'] = X.shape[1]                                       # number of coefficients
mydata['Kc'] = X.shape[1

â€‹

# Fit
stan_code = """
data{
int N;
int Kb;
int Kc;
matrix[N, Kb] Xb;
matrix[N, Kc] Xc;
int Y[N];
}
parameters{
vector[Kc] beta;
vector[Kb] gamma;
real<lower=0> alpha;

}
transformed parameters{
vector[N] mu;
vector[N] Pi;

mu = exp(Xc * beta);
for (i in 1:N) Pi[i] = inv_logit(Xb[i] * gamma);
}
model{
vector[N] LL;

for (i in 1:N) {
if (Y[i] == 0) {
LL[i] = log_sum_exp(bernoulli_lpmf(1| Pi[i]),
bernoulli_lpmf(0| Pi[i]) +
neg_binomial_2_lpmf(Y[i]| mu[i], 1/alpha));
} else {
LL[i] = bernoulli_lpmf(0| Pi[i]) +
neg_binomial_2_lpmf(Y[i]| mu[i], 1/alpha);
}
}
target += LL;
}
"""

â€‹

# Run mcmc
fit = pystan.stan(model_code=stan_code, data=mydata, iter=7000, chains=3,
warmup=3500, n_jobs=3)

â€‹

# Output
nlines = 12                                                           # number of lines in screen output

â€‹

output = str(fit).split('\n')
for item in output[:nlines]:
print(item)

=================================================================

Output on screen:

â€‹

Inference for Stan model: anon_model_9dd78dbc1f3c1ddb1d2ba6cf92010d52.
3 chains, each with iter=7000; warmup=3500; thin=1;
post-warmup draws per chain=3500, total post-warmup draws=10500.

â€‹

mean     se_mean          sd       2.5%      25%       50%      75%     97.5%      n_eff     Rhat
beta[0]            1.16         1.3e-3         0.1       0.97       1.09       1.16      1.22       1.35      5372.0        1.0
beta[1]            1.87         1.8e-3       0.13       1.61       1.78       1.87      1.96       2.13      5433.0        1.0
beta[2]            1.36         7.6e-4       0.07       1.22       1.31       1.35        1.4       1.49      7996.0        1.0
gamma[0]       2.09         1.2e-3         0.1         1.9       2.03       2.09      2.15       2.28      6212.0        1.0
gamma[1]      -4.97         2.5e-3       0.19      -5.33     -5.09      -4.97    -4.84      -4.61      5331.0        1.0
gamma[2]       2.94         1.5e-3       0.11        2.73      2.87       2.94      3.02       3.16      5371.0        1.0
alpha              1.85          1.2e-3       0.09       1.67       1.78       1.85      1.91       2.04      6520.0        1.0

bottom of page