def test_merge_results(self):
results1 = load_results('../data/1000 runs scarcity.tar.gz')
results2 = load_results('../data/1000 runs scarcity.tar.gz')
n1 = results1[0].shape[0]
n2 = results2[0].shape[0]
merged = merge_results(results1, results2)
self.assertEqual(merged[0].shape[0], n1 + n2)
def test_merge_results(self):
results1 = load_results('../data/1000 runs scarcity.tar.gz')
results2 = load_results('../data/1000 runs scarcity.tar.gz')
n1 = results1[0].shape[0]
n2 = results2[0].shape[0]
merged = merge_results(results1, results2)
self.assertEqual(merged[0].shape[0], n1+n2)
def test_load_results(self):
# test for 1d
# test for 2d
# test for 3d
nr_experiments = 10000
experiments = pd.DataFrame(index=np.arange(nr_experiments),
columns={
'x': np.float,
'y': np.float
})
experiments['x'] = np.random.rand(nr_experiments)
experiments['y'] = np.random.rand(nr_experiments)
outcome_a = np.zeros((nr_experiments, 1))
results = (experiments, {'a': outcome_a})
save_results(results, '../data/test.tar.gz')
loaded_experiments, outcomes = load_results('../data/test.tar.gz')
self.assertTrue(np.all(np.allclose(outcomes['a'], outcome_a)))
self.assertTrue(
np.all(np.allclose(experiments['x'], loaded_experiments['x'])))
self.assertTrue(
np.all(np.allclose(experiments['y'], loaded_experiments['y'])))
os.remove('../data/test.tar.gz')
nr_experiments = 1000
nr_timesteps = 100
nr_replications = 10
experiments = pd.DataFrame(index=np.arange(nr_experiments),
columns={
'x': np.float,
'y': np.float
})
experiments['x'] = np.random.rand(nr_experiments)
experiments['y'] = np.random.rand(nr_experiments)
outcome_a = np.zeros((nr_experiments, nr_timesteps, nr_replications))
results = (experiments, {'a': outcome_a})
save_results(results, '../data/test.tar.gz')
loaded_experiments, outcomes = load_results('../data/test.tar.gz')
os.remove('../data/test.tar.gz')
self.assertTrue(np.all(np.allclose(outcomes['a'], outcome_a)))
self.assertTrue(
np.all(np.allclose(experiments['x'], loaded_experiments['x'])))
self.assertTrue(
np.all(np.allclose(experiments['y'], loaded_experiments['y'])))
def test_load_results(self):
# test for 1d
# test for 2d
# test for 3d
nr_experiments = 10000
experiments = np.recarray((nr_experiments, ),
dtype=[('x', np.float), ('y', np.float)])
experiments['x'] = np.random.rand(nr_experiments)
experiments['y'] = np.random.rand(nr_experiments)
outcome_a = np.zeros((nr_experiments, 1))
results = (experiments, {'a': outcome_a})
save_results(results, '../data/test.tar.gz')
loaded_experiments, outcomes = load_results('../data/test.tar.gz')
self.assertTrue(np.all(np.allclose(outcomes['a'], outcome_a)))
self.assertTrue(
np.all(np.allclose(experiments['x'], loaded_experiments['x'])))
self.assertTrue(
np.all(np.allclose(experiments['y'], loaded_experiments['y'])))
os.remove('../data/test.tar.gz')
nr_experiments = 1000
nr_timesteps = 100
nr_replications = 10
experiments = np.recarray((nr_experiments, ),
dtype=[('x', np.float), ('y', np.float)])
experiments['x'] = np.random.rand(nr_experiments)
experiments['y'] = np.random.rand(nr_experiments)
outcome_a = np.zeros((nr_experiments, nr_timesteps, nr_replications))
results = (experiments, {'a': outcome_a})
save_results(results, '../data/test.tar.gz')
loaded_experiments, outcomes = load_results('../data/test.tar.gz')
os.remove('../data/test.tar.gz')
self.assertTrue(np.all(np.allclose(outcomes['a'], outcome_a)))
self.assertTrue(
np.all(np.allclose(experiments['x'], loaded_experiments['x'])))
self.assertTrue(
np.all(np.allclose(experiments['y'], loaded_experiments['y'])))
def test_load_results(self):
# test for 1d
# test for 2d
# test for 3d
nr_experiments = 10000
experiments = pd.DataFrame(index=np.arange(nr_experiments),
columns={'x':np.float,
'y': np.float})
experiments['x'] = np.random.rand(nr_experiments)
experiments['y'] = np.random.rand(nr_experiments)
outcome_a = np.zeros((nr_experiments,1))
results = (experiments, {'a': outcome_a})
save_results(results, '../data/test.tar.gz')
loaded_experiments, outcomes = load_results('../data/test.tar.gz')
self.assertTrue(np.all(np.allclose(outcomes['a'],outcome_a)))
self.assertTrue(np.all(np.allclose(experiments['x'],loaded_experiments['x'])))
self.assertTrue(np.all(np.allclose(experiments['y'],loaded_experiments['y'])))
os.remove('../data/test.tar.gz')
nr_experiments = 1000
nr_timesteps = 100
nr_replications = 10
experiments = pd.DataFrame(index=np.arange(nr_experiments),
columns={'x':np.float,
'y': np.float})
experiments['x'] = np.random.rand(nr_experiments)
experiments['y'] = np.random.rand(nr_experiments)
outcome_a = np.zeros((nr_experiments,nr_timesteps,nr_replications))
results = (experiments, {'a': outcome_a})
save_results(results, '../data/test.tar.gz')
loaded_experiments, outcomes = load_results('../data/test.tar.gz')
os.remove('../data/test.tar.gz')
self.assertTrue(np.all(np.allclose(outcomes['a'],outcome_a)))
self.assertTrue(np.all(np.allclose(experiments['x'],loaded_experiments['x'])))
self.assertTrue(np.all(np.allclose(experiments['y'],loaded_experiments['y'])))
def test_load_results(self):
# test for 1d
# test for 2d
# test for 3d
nr_experiments = 10000
experiments = np.recarray((nr_experiments, ),
dtype=[('x', float), ('y', float)])
outcome_a = np.zeros((nr_experiments, 1))
results = (experiments, {'a': outcome_a})
save_results(results, u'../data/test.tar.gz')
experiments, outcomes = load_results(u'../data/test.tar.gz')
logical = np.allclose(outcomes['a'], outcome_a)
os.remove('../data/test.tar.gz')
# if logical:
# ema_logging.info('1d loaded successfully')
nr_experiments = 1000
nr_timesteps = 100
nr_replications = 10
experiments = np.recarray((nr_experiments, ),
dtype=[('x', float), ('y', float)])
outcome_a = np.zeros((nr_experiments, nr_timesteps, nr_replications))
results = (experiments, {'a': outcome_a})
save_results(results, '../data/test.tar.gz')
experiments, outcomes = load_results('../data/test.tar.gz')
logical = np.allclose(outcomes['a'], outcome_a)
os.remove('../data/test.tar.gz')
def test_load_results(self):
# test for 1d
# test for 2d
# test for 3d
nr_experiments = 10000
experiments = np.recarray((nr_experiments,),
dtype=[('x', float), ('y', float)])
outcome_a = np.zeros((nr_experiments,1))
results = (experiments, {'a': outcome_a})
save_results(results, u'../data/test.tar.gz')
experiments, outcomes = load_results(u'../data/test.tar.gz')
logical = np.allclose(outcomes['a'],outcome_a)
os.remove('../data/test.tar.gz')
# if logical:
# ema_logging.info('1d loaded successfully')
nr_experiments = 1000
nr_timesteps = 100
nr_replications = 10
experiments = np.recarray((nr_experiments,),
dtype=[('x', float), ('y', float)])
outcome_a = np.zeros((nr_experiments,nr_timesteps,nr_replications))
results = (experiments, {'a': outcome_a})
save_results(results, u'../data/test.tar.gz')
experiments, outcomes = load_results(u'../data/test.tar.gz')
logical = np.allclose(outcomes['a'],outcome_a)
os.remove('../data/test.tar.gz')
def periodDominance(ds):
Y = np.fft.rfft(ds)
n = len(Y)
powerSpect = np.abs(Y)**2
timeStep = 1
freq = np.fft.fftfreq(n, d=timeStep)
print len(freq), len(powerSpect)
for i in range(len(freq) / 2 + 1):
print freq[i], 1 / freq[i], powerSpect[i]
if __name__ == '__main__':
cases, results = utilities.load_results('PatternSet_Periodic.cpickle')
dataSeries = results.get('outcome')
ds1 = dataSeries[25]
ds2 = dataSeries[26]
print linearFit(ds1)
print quadraticFit(ds1)
print mean(ds1), variance(ds1), stdDev(ds1)
print autoCovariance(ds1, 0)
for k in range(31):
print k, autoCorrelation(ds1, k)
for k in range(31):
print k, crossCorrelation(ds1, ds2, k)
periodDominance(ds1)
return var
def periodDominance(ds):
Y = np.fft.rfft(ds)
n = len(Y)
powerSpect = np.abs(Y)**2
timeStep = 1
freq = np.fft.fftfreq(n, d=timeStep)
print(len(freq), len(powerSpect))
for i in range(len(freq)/2+1):
print(freq[i], 1/freq[i], powerSpect[i])
if __name__ == '__main__':
cases, results = utilities.load_results('PatternSet_Periodic.cpickle')
dataSeries = results.get('outcome')
ds1 = dataSeries[25]
ds2 = dataSeries[26]
print(linearFit(ds1))
print(quadraticFit(ds1))
print(mean(ds1), variance(ds1), stdDev(ds1))
print(autoCovariance(ds1,0))
for k in range(31):
print(k,autoCorrelation(ds1,k))
for k in range(31):
print(k, crossCorrelation(ds1,ds2,k))
periodDominance(ds1)
Scenario, Policy, Constraint)
from ema_workbench import load_results
from ema_workbench.analysis import prim, dimensional_stacking, cart
from ema_workbench.util import ema_logging, utilities
from ema_workbench.em_framework.optimization import (HyperVolume,
EpsilonProgress)
from ema_workbench.em_framework import sample_uncertainties, MonteCarloSampler
from ema_workbench.em_framework.evaluators import BaseEvaluator
ema_logging.log_to_stderr(ema_logging.INFO)
from dike_model_function import DikeNetwork # @UnresolvedImport
from problem_formulation import get_model_for_problem_formulation
results = utilities.load_results('results/base_case_75.csv')
experiments, outcomes = results
outcomes = pd.DataFrame(outcomes)
experiments = pd.DataFrame(experiments)
outcome_total = pd.read_csv('results/75policies_with_total_values.csv')
results = experiments.join(outcome_total)
results = results.drop(columns="model")
# minimise the worst outcome and minimise the standard deviation
#defined as the median value multiplied by the interquartile distance plus one
def robustness(result):
def aggregate_outcomes(results, outcome):
list_outcomes_columns = []
for i in results.columns:
if outcome in i:
list_outcomes_columns.append(i)
results["Total " +
str(outcome)] = results[list_outcomes_columns].sum(axis=1)
# ### Find the ranges for epsilon and hypervolume convergence
#
# To set $\epsilon$ values, we must minimize noise by first running a robust optimize quickly to see a Pareto front develop as stated in section 3.4 of doi: 10.1016/j.envsoft.2011.04.003 (we don't only look at Monte Carlo policies in hope that this will save time).
results = utilities.load_results('Outcomes/MOROpolicies50Scenarios.csv')
experiments, outcomes = results
outcomes = pd.DataFrame(outcomes)
experiments = pd.DataFrame(experiments)
results = experiments.join(outcomes)
results = results.drop(columns="model")
# Aggregate
aggregate_outcomes(outcomes, "Expected Annual Damage")
aggregate_outcomes(outcomes, "Dike Investment Costs")
aggregate_outcomes(outcomes, "Expected Number of Deaths")
aggregate_outcomes(outcomes, "RfR Total Costs")
aggregate_outcomes(outcomes, "Expected Evacuation Costs")
