def grid_search_cv(pipeline, chain_names, chain_hyperparameter_space, dataset_name, verbose=0):
tuned_parameters = chain_hyperparameter_space.get_grid_parameters()
if verbose > 0:
print("Parameters : values to test")
for k in tuned_parameters:
print(" --", k, ":", tuned_parameters[k])
"""
Grid search from scikit-learn:
GridSearchCV(estimator, param_grid, scoring=None, fit_params=None,
n_jobs=1, iid=True, refit=True, cv=None, verbose=0,
pre_dispatch=‘2*n_jobs’, error_score=’raise’, return_train_score=’warn’)
param_grid : dict or list of dictionaries
Dictionary with parameters names (string) as keys and lists of parameter
settings to try as values, or a list of such dictionaries, in which case
the grids spanned by each dictionary in the list are explored. This
enables searching over any sequence of parameter settings.
"""
model = grid_search.GridSearch(pipeline, chain_names, tuned_parameters,
dataset_name, verbose=verbose)
return model
def test_4(self):
"""
Chromosome: sequence of clients separed by an 'X' when a new vehicle is assigned
Selection: roulette-wheel
Crossover operator: simple random crossover
Mutation operator: simple random mutation
Elitism is enabled
Termination criteria: number of generations = 100
Parameters:
population_size: 97
reproduction rate: 0.25
crossover rate: 0.625
mutation rate: 0.125
"""
fname = './input/A-n80-k10.vrp'
nodes, capacity, distances, demand = self.load_test(fname)
individual_factory = cvrp.CVRPIndividualFactory(
nodes, capacity, distances, demand, individual_type='corrected')
termination_criteria = ga.NumberOfGenerationsTerminationCriteria(
number_of_generations=100)
solver = ga.GeneticAlgorithm(individual_factory,
population_size=97,
reproduction=0.25,
crossover=0.625,
mutation=0.125,
elitism=True,
termination_criteria=termination_criteria)
if self.grid_search:
params = {
"population_size":
numpy.logspace(3, 12, base=2, num=6, dtype=int),
"operators_rate":
filter(
lambda x: sum(x) == 1.0,
itertools.product(numpy.arange(.125, 0.875, .125),
repeat=3)),
"elitism": [True],
"termination_criteria": [
ga.NumberOfGenerationsTerminationCriteria(
number_of_generations=100)
]
}
grid = grid_search.GridSearch(solver, params)
grid.search(0.0)
grid_scores = grid.get_grid_scores()
fname = './results/correction_operator/A-n80-k10.vrp.grid.csv'
grid_scores.to_csv(fname, sep=',', index=False)
sys.stdout.write(
"Finished. Results are at: ./results/correction_operator/A-n80-k10.vrp.grid.csv\n"
)
else:
sys.stdout.write(
"Starting test_4: CORRECTION + SIMPLE RANDOM OPERATORS, ELITISM ENABLED\n"
)
sys.stdout.write("Input: ./tests/vrp/A-n80-k10.vrp\n")
solver.init_population()
solver.evolve()
info = solver.get_generation_info()
fname = './results/correction_operator/A-n80-k10.vrp.csv'
info.to_csv(fname, sep=',', index=False)
plt.plot(info['generation'],
info['min'],
"r",
label="melhor",
linewidth=2)
plt.plot(info['generation'],
info['mean'],
"b",
label="media",
linewidth=2)
plt.plot(info['generation'], info['std'], "k.", label="desvio")
legend = plt.legend(loc='lower right', numpoints=1)
plt.xlabel("geracoes")
plt.ylabel("fitness")
plt.yscale('log')
plt.show()
sys.stdout.write(
"Finished. Results are at: ./results/correction_operator/A-n80-k10.vrp.csv\n"
)
assert True