def test_bootstrap(self):
objval, thetavals = self.pest.theta_est()
num_bootstraps = 10
theta_est = self.pest.bootstrap(num_bootstraps)
num_samples = theta_est['samples'].apply(len)
self.assertTrue(len(theta_est.index), 10)
self.assertTrue(num_samples.equals(pd.Series([6] * 10)))
filename = os.path.abspath(os.path.join(testdir, 'pairwise.png'))
if os.path.isfile(filename):
os.remove(filename)
graphics.pairwise_plot(theta_est, filename=filename)
#self.assertTrue(os.path.isfile(filename))
filename = os.path.abspath(
os.path.join(testdir, 'pairwise_bootstrap.png'))
if os.path.isfile(filename):
os.remove(filename)
graphics.pairwise_bootstrap_plot(theta_est,
thetavals,
0.8,
filename=filename)
fsfilename = "examples.contrib.projects.mea_simple.parameter_estimate.mea_estimate_pysp"
pest = parmest.ParmEstimator(fsfilename, "pysp_instance_creation_callback",
"SecondStageCost", range(1, S + 1), thetalist)
mpii = mpiu.MPIInterface() # works with, or without, mpi
objval, thetavals = pest.theta_est()
if mpii.rank in [0, None]:
print("objective value=", str(objval))
for t in thetavals:
print(t, "=", thetavals[t])
print("====")
### Parameter estimation with bootstrap
np.random.seed(1134) # make it reproducible
num_bootstraps = 10
bootstrap_theta = pest.bootstrap(num_bootstraps)
if mpii.rank in [0, None]:
print("Bootstrap:")
print(bootstrap_theta)
grph.pairwise_bootstrap_plot(bootstrap_theta,
thetavals,
0.8,
filename="MEA_boot.png")
### Make the LR plot
# SEE = pest.likelihood_ratio(0.8)
# grph.pairwise_likelihood_ratio_plot(SSE, objval, 0.8, S, "MEA_LR.png")
objval, thetavals = pest.theta_est()
if mpii.rank in [0, None]:
print("objective value=", str(objval))
print("theta-star=", str(thetavals))
### Parameter estimation with bootstrap
alpha = 0.8
num_bootstraps = 10
bootstrap_theta = pest.bootstrap(num_bootstraps)
if mpii.rank in [0, None]:
print("Bootstrap:")
print(bootstrap_theta)
grph.pairwise_plot(bootstrap_theta, filename="RB.png")
grph.pairwise_bootstrap_plot(bootstrap_theta,
thetavals,
alpha,
filename="RB_boot.png")
### Likelihood ratio
alpha = 0.8
search_ranges = {}
search_ranges["asymptote"] = np.arange(10, 30, 2) # np.arange(10, 30, 0.01)
search_ranges["rate_constant"] = np.arange(0, 1.5,
0.1) # np.arange(0, 1.5, 0.005)
SSE = pest.likelihood_ratio(search_ranges=search_ranges)
if mpii.rank in [0, None]:
print("Likelihood Ratio:")
print(SSE)
grph.pairwise_likelihood_ratio_plot(SSE,
objval,
alpha,