def test_json_gzip_dependent_variable():
from EvoDAG.command_line import params
import gzip
import tempfile
import json
import os
os.environ['KLASS'] = 'dependent'
fname = tempfile.mktemp() + '.gz'
with gzip.open(fname, 'wb') as fpt:
for x, y in zip(X, cl):
a = {k: v for k, v in enumerate(x)}
a['dependent'] = int(y)
a['num_terms'] = len(x)
try:
fpt.write(bytes(json.dumps(a) + '\n', encoding='utf-8'))
except TypeError:
fpt.write(json.dumps(a) + '\n')
print("termine con el json")
sys.argv = [
'EvoDAG', '-C', '-Poutput.evodag', '--json', '-e1', '-p3', '-r2', fname
]
params()
os.unlink(fname)
print(open('output.evodag').read())
os.unlink('output.evodag')
default_nargs()
del os.environ['KLASS']
def test_number_multiple_outputs_regression():
import os
from EvoDAG.command_line import params, train, predict
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-R', '--parameters', 'cache.evodag',
'-p3', '-e1', '-r2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--output-dim=3',
'--model', 'model.evodag', '--test', fname, fname
]
train()
sys.argv = [
'EvoDAG', '--output', 'output.evodag', '--decision-function',
'--model', 'model.evodag', fname
]
predict()
os.unlink(fname)
os.unlink('cache.evodag')
os.unlink('model.evodag')
l = open('output.evodag').readline()
os.unlink('output.evodag')
default_nargs()
assert len(l.split(',')) == 3
def test_utils_graphviz_terminals():
import os
from EvoDAG.command_line import params, train, utils
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '--parameters', 'cache.evodag', '-p3', '-e1', '-r2',
fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--model',
'model.evodag', '--test', fname, fname
]
train()
sys.argv = [
'EvoDAG', '--output', 'output.evodag', '--decision-function', '-u2',
'--model', 'model.evodag', fname
]
sys.argv = [
'EvoDAG', '-G', '-oevodag_gv', '--remove-terminals', 'model.evodag'
]
utils()
os.unlink(fname)
os.unlink('cache.evodag')
os.unlink('model.evodag')
for i in range(2):
os.unlink('evodag_gv/evodag-%s' % i)
os.rmdir('evodag_gv')
default_nargs()
def test_raw_outputs_classification():
import os
from EvoDAG.command_line import params, train, predict
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-C', '--parameters', 'cache.evodag',
'-p3', '-e1', '-r2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--output-dim=3',
'--model', 'model.evodag', '--test', fname, fname
]
train()
sys.argv = [
'EvoDAG', '--output', 'output.evodag', '--raw-outputs', '--model',
'model.evodag', fname
]
predict()
os.unlink(fname)
os.unlink('cache.evodag')
os.unlink('model.evodag')
l = open('output.evodag').readline()
os.unlink('output.evodag')
default_nargs()
assert len(l.split(',')) == 6
def test_multiple_outputs():
from EvoDAG.command_line import params
import gzip
import tempfile
import json
fname = tempfile.mktemp() + '.gz'
with gzip.open(fname, 'wb') as fpt:
for x, y in zip(X, cl):
a = {k: v for k, v in enumerate(x)}
a['klass'] = int(y)
a['num_terms'] = len(x)
try:
fpt.write(bytes(json.dumps(a) + '\n', encoding='utf-8'))
except TypeError:
fpt.write(json.dumps(a) + '\n')
print("termine con el json")
sys.argv = [
'EvoDAG', '-C', '-Poutput.evodag', '--json', '-e1', '-p3', '-r2', fname
]
params()
os.unlink(fname)
d = json.loads(open('output.evodag').read())
assert d[0]['multiple_outputs']
os.unlink('output.evodag')
default_nargs()
def test_create_ensemble():
import os
from EvoDAG.command_line import params, train, utils
from EvoDAG.model import Ensemble
import pickle
import gzip
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-R', '--parameters', 'cache.evodag',
'-p3', '-e2', '-r2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n1', '--output-dim=3',
'--seed=1', '--model', 'model.evodag.1', '--test', fname, fname
]
train()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n1', '--output-dim=3',
'--seed=2', '--model', 'model.evodag.2', '--test', fname, fname
]
train()
os.unlink('cache.evodag')
sys.argv = [
'EvoDAG', '--create-ensemble', '-omodel.evodag', '-n3',
'model.evodag.* model.evodag.1'
]
utils()
os.unlink('model.evodag.1')
os.unlink('model.evodag.2')
with gzip.open('model.evodag', 'r') as fpt:
ens = pickle.load(fpt)
assert isinstance(ens, Ensemble)
os.unlink('model.evodag')
default_nargs()
def test_utils_graphviz():
import os
from EvoDAG.command_line import params, train, utils
fname = training_set()
sys.argv = ['EvoDAG', '-C', '--parameters',
'cache.evodag', '-p3', '-e1',
'-r2', fname]
params()
sys.argv = ['EvoDAG', '--parameters', 'cache.evodag',
'-n2',
'--model', 'model.evodag',
'--test', fname, fname]
train()
sys.argv = ['EvoDAG', '--output', 'output.evodag',
'--decision-function',
'-u2',
'--model', 'model.evodag', fname]
sys.argv = ['EvoDAG', '-G', '-oevodag_gv',
'model.evodag']
utils()
os.unlink(fname)
os.unlink('cache.evodag')
os.unlink('model.evodag')
for i in range(2):
for j in range(3):
os.unlink('evodag_gv/evodag-%s-%s.gv' % (i, j))
os.rmdir('evodag_gv')
default_nargs()
def test_logging():
from EvoDAG.command_line import params
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-R', '--parameters', 'cache.evodag',
'--verbose=0', '-p3', '-e2', '-r2', fname
]
params()
def test_cpu_cores():
import os
from EvoDAG.command_line import params
fname = training_set()
sys.argv = ['EvoDAG', '-C', '-u2', '--parameters',
'cache.evodag.gz', '-p3', '-e2', '-r', '2', fname]
params()
assert os.path.isfile('cache.evodag.gz')
os.unlink('cache.evodag.gz')
os.unlink(fname)
def test_cpu_cores():
import os
from EvoDAG.command_line import params
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '-u2', '--parameters', 'cache.evodag.gz', '-p3', '-e2',
'-r', '2', fname
]
params()
assert os.path.isfile('cache.evodag.gz')
os.unlink('cache.evodag.gz')
os.unlink(fname)
default_nargs()
def test_get_best_params_files():
import numpy as np
import gzip
import pickle
import os
from EvoDAG.command_line import params, train, utils
import shutil
from glob import glob
if os.path.isdir('cache'):
shutil.rmtree('cache')
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '-Pcache', '--only-paramsfiles', '-r', '2', fname
]
params()
for p in glob(os.path.join('cache', '*_params.json')):
basename = p.split('_params.json')[0]
for s in range(3):
model = basename + '_%s.model' % s
sys.argv = [
'EvoDAG',
'-s%s' % s,
'-P%s' % p,
'-m%s' % model, '-n1', fname
]
train()
R = []
for p in range(2):
l = []
for s in range(3):
with gzip.open(os.path.join('cache',
'%s_%s.model' % (p, s))) as fpt:
m = pickle.load(fpt)
l.append(m.fitness_vs * -1)
R.append((p, l))
print(R)
m = np.array([np.median(x[1]) for x in R])
m = m - np.min(m)
param = ['%s_params.json' % k for k, x in enumerate(m) if x == 0]
sys.argv = ['EvoDAG', '--best-params-file', 'cache']
c = utils(output=True)
print(c.best_params, param)
assert c.best_params in param
sys.argv = ['EvoDAG', '-u2', '--best-params-file', 'cache']
c = utils(output=True)
print(c.best_params, param, '*')
assert c.best_params in param
if os.path.isdir('cache'):
shutil.rmtree('cache')
default_nargs()
def test_params_files():
import os
from EvoDAG.command_line import params
import shutil
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '-Pcache', '--only-paramsfiles', '-r', '2', fname
]
params()
if os.path.isdir('cache'):
shutil.rmtree('cache')
default_nargs()
return
assert False
def test_utils_params_stats():
from EvoDAG.command_line import params, utils
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-R', '--parameters', 'cache.evodag',
'-p3', '-e1', '-r3', fname
]
params()
sys.argv = ['EvoDAG', '-P', '-o', fname, 'cache.evodag']
utils()
with open(fname) as fpt:
print(fpt.read())
os.unlink(fname)
os.unlink('cache.evodag')
default_nargs()
def test_word2id():
import tempfile
from EvoDAG.command_line import params
fname = tempfile.mktemp()
id2word = dict([[0, 'a'], [1, 'b'], [2, 'c']])
with open(fname, 'w') as fpt:
for x, v in zip(X, cl):
l = x.tolist()
l.append(id2word[v])
fpt.write(','.join(map(str, l)))
fpt.write('\n')
sys.argv = ['EvoDAG', '-C', '-Pparams.gz', '-r2', '-e2', '-p3', fname]
params()
os.unlink(fname)
os.unlink('params.gz')
def test_classifier_params():
import os
import json
from EvoDAG.command_line import params
fname = training_set()
sys.argv = ['EvoDAG', '-C', '--parameters',
'cache.evodag', '-p3', '-e1',
'-r', '2', fname]
params()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'classifier' in a
assert a['classifier']
os.unlink('cache.evodag')
def test_random_generations():
import os
import json
from EvoDAG.command_line import params
fname = training_set()
sys.argv = ['EvoDAG', '-C', '--parameters',
'cache.evodag', '-p3', '-e2',
'--random-generations', '1',
'-r', '2', fname]
params()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'random_generations' in a
os.unlink('cache.evodag')
def test_word2id():
import tempfile
from EvoDAG.command_line import params
fname = tempfile.mktemp()
id2word = dict([[0, 'a'], [1, 'b'], [2, 'c']])
with open(fname, 'w') as fpt:
for x, v in zip(X, cl):
l = x.tolist()
l.append(id2word[v])
fpt.write(','.join(map(str, l)))
fpt.write('\n')
sys.argv = ['EvoDAG', '-C', '-Pparams.gz', '-r2', '-e2', '-p3', fname]
params()
os.unlink(fname)
os.unlink('params.gz')
default_nargs()
def test_regressor_params():
import os
import json
from EvoDAG.command_line import params
fname = training_set()
sys.argv = ['EvoDAG', '-R', '--parameters',
'cache.evodag', '-p3', '-e1',
'-r', '2', fname]
params()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'classifier' in a
assert not a['classifier']
assert a['popsize'] == 3
assert a['early_stopping_rounds'] == 1
os.unlink('cache.evodag')
def test_parameters_values():
import os
from EvoDAG.command_line import params
import json
fname = training_set()
with open('p.conf', 'w') as fpt:
fpt.write(json.dumps(dict(popsize=['x'])))
sys.argv = ['EvoDAG', '-C', '-Pcache.evodag.gz', '-p3', '-e2',
'--parameters-values', 'p.conf',
'-r', '2', fname]
try:
params()
assert False
except ValueError:
pass
os.unlink('p.conf')
os.unlink(fname)
def test_classifier_params():
import os
import json
from EvoDAG.command_line import params
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '--parameters', 'cache.evodag', '-p3', '-e1', '-r',
'2', fname
]
params()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'classifier' in a
assert a['classifier']
os.unlink('cache.evodag')
default_nargs()
def test_model_seed():
import os
from EvoDAG.command_line import params, train
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-R', '--parameters', 'cache.evodag',
'-p3', '-e2', '-r2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--output-dim=3',
'--seed=1', '--model', 'model.evodag', '--test', fname, fname
]
c = train(output=True)
os.unlink('cache.evodag')
default_nargs()
assert c.data.seed == 1
def test_get_best_params_files():
import numpy as np
import gzip
import pickle
import os
from EvoDAG.command_line import params, train, utils
import shutil
from glob import glob
if os.path.isdir('cache'):
shutil.rmtree('cache')
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '-Pcache', '--only-paramsfiles', '-r', '2', fname
]
params()
for p in glob('cache/*_params.json'):
basename = p.split('_params.json')[0]
for s in range(3):
model = basename + '_%s.model' % s
sys.argv = [
'EvoDAG',
'-s%s' % s,
'-P%s' % p,
'-m%s' % model, '-n1', fname
]
train()
R = []
for p in range(2):
l = []
for s in range(3):
with gzip.open('cache/%s_%s.model' % (p, s)) as fpt:
m = pickle.load(fpt)
l.append(m.fitness_vs * -1)
R.append((p, l))
m = min(R, key=lambda x: np.median(x[1]))
param = '%s_params.json' % m[0]
sys.argv = ['EvoDAG', '--best-params-file', 'cache']
c = utils(output=True)
assert c.best_params == param
sys.argv = ['EvoDAG', '-u2', '--best-params-file', 'cache']
c = utils(output=True)
assert c.best_params == param
if os.path.isdir('cache'):
shutil.rmtree('cache')
default_nargs()
def test_json():
from EvoDAG.command_line import params
import tempfile
import json
fname = tempfile.mktemp()
with open(fname, 'w') as fpt:
for x, y in zip(X, cl):
a = {k: v for k, v in enumerate(x)}
a['klass'] = int(y)
a['num_terms'] = len(x)
fpt.write(json.dumps(a) + '\n')
print("termine con el json")
sys.argv = ['EvoDAG', '-C', '-Poutput.evodag', '--json',
'-e1', '-p3', '-r2', fname]
params()
os.unlink(fname)
print(open('output.evodag').read())
os.unlink('output.evodag')
def test_train():
import os
from EvoDAG.command_line import params, CommandLineTrain
fname = training_set()
sys.argv = ['EvoDAG', '-C', '-Pcache.evodag.gz',
'-p3', '-e2', '-r2', fname]
params()
sys.argv = ['EvoDAG', '-Pcache.evodag.gz',
'-n2',
'--model', 'model.evodag',
'--test_set', fname, fname]
c = CommandLineTrain()
c.parse_args()
assert os.path.isfile(c.data.test_set)
assert os.path.isfile('model.evodag')
os.unlink(fname)
os.unlink('cache.evodag.gz')
os.unlink('model.evodag')
default_nargs()
def test_parameters_values():
import os
from EvoDAG.command_line import params
import json
fname = training_set()
with open('p.conf', 'w') as fpt:
fpt.write(json.dumps(dict(popsize=['x'], multiple_outputs=[True])))
sys.argv = [
'EvoDAG', '-C', '-Pcache.evodag.gz', '-p3', '-e2',
'--parameters-values', 'p.conf', '-r', '2', fname
]
try:
params()
assert False
except ValueError:
pass
os.unlink('p.conf')
os.unlink(fname)
default_nargs()
def test_regressor_params():
import os
import json
from EvoDAG.command_line import params
fname = training_set()
sys.argv = [
'EvoDAG', '-R', '--parameters', 'cache.evodag', '-p3', '-e1', '-r',
'2', fname
]
params()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'classifier' in a
assert not a['classifier']
assert a['popsize'] == 3
assert a['early_stopping_rounds'] == 1
os.unlink('cache.evodag')
default_nargs()
def test_model_used_inputs_number():
import os
from EvoDAG.command_line import params, train, utils
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-R', '--parameters', 'cache.evodag',
'-p3', '-e1', '-r2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--output-dim=3',
'--model', 'model.evodag', '--test', fname, fname
]
train()
sys.argv = ['EvoDAG', '--used-inputs-number', 'model.evodag']
utils()
os.unlink('cache.evodag')
os.unlink('model.evodag')
default_nargs()
def test_train():
import os
from EvoDAG.command_line import params, CommandLineTrain
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '-Pcache.evodag.gz', '-p3', '-e2', '-r2', fname
]
params()
sys.argv = [
'EvoDAG', '-Pcache.evodag.gz', '-n2', '--model', 'model.evodag',
'--test_set', fname, fname
]
c = CommandLineTrain()
c.parse_args()
assert os.path.isfile(c.data.test_set)
assert os.path.isfile('model.evodag')
os.unlink(fname)
os.unlink('cache.evodag.gz')
os.unlink('model.evodag')
default_nargs()
def test_json():
from EvoDAG.command_line import params
import tempfile
import json
fname = tempfile.mktemp()
with open(fname, 'w') as fpt:
for x, y in zip(X, cl):
a = {k: v for k, v in enumerate(x)}
a['klass'] = int(y)
a['num_terms'] = len(x)
fpt.write(json.dumps(a) + '\n')
print("termine con el json")
sys.argv = [
'EvoDAG', '-C', '-Poutput.evodag', '--json', '-e1', '-p3', '-r2', fname
]
params()
os.unlink(fname)
print(open('output.evodag').read())
os.unlink('output.evodag')
default_nargs()
def test_params():
import os
import gzip
from EvoDAG.command_line import params
import json
fname = training_set()
sys.argv = ['EvoDAG', '-C', '--parameters',
'cache.evodag.gz', '-p3', '-e2', '-r', '2', fname]
params()
os.unlink(fname)
assert os.path.isfile('cache.evodag.gz')
with gzip.open('cache.evodag.gz', 'rb') as fpt:
try:
d = fpt.read()
a = json.loads(str(d, encoding='utf-8'))
except TypeError:
a = json.loads(d)
os.unlink('cache.evodag.gz')
assert len(a) == len([x for x in a if 'fitness' in x])
print(a)
def test_predict():
import os
from EvoDAG.command_line import params, train, predict
fname = training_set()
sys.argv = ['EvoDAG', '-C', '--parameters',
'cache.evodag.gz', '-p3', '-e2', '-r', '2', fname]
params()
sys.argv = ['EvoDAG', '--parameters', 'cache.evodag.gz',
'-n2',
'--model', 'model.evodag',
'--test', fname, fname]
train()
sys.argv = ['EvoDAG', '--output', 'output.evodag',
'--model', 'model.evodag', fname]
predict()
os.unlink(fname)
os.unlink('cache.evodag.gz')
os.unlink('model.evodag')
assert os.path.isfile('output.evodag')
os.unlink('output.evodag')
default_nargs()
def test_decision_function():
import os
from EvoDAG.command_line import params, train, predict
fname = training_set()
sys.argv = ['EvoDAG', '-C', '--parameters',
'cache.evodag', '-p3', '-e1',
'-r', '1', fname]
params()
sys.argv = ['EvoDAG', '--parameters', 'cache.evodag',
'-n2',
'--model', 'model.evodag',
'--test', fname, fname]
train()
sys.argv = ['EvoDAG', '--output', 'output.evodag',
'--decision-function',
'--model', 'model.evodag', fname]
predict()
os.unlink(fname)
os.unlink('cache.evodag')
os.unlink('model.evodag')
os.unlink('output.evodag')
default_nargs()
def test_json_gzip():
from EvoDAG.command_line import params
import gzip
import tempfile
import json
fname = tempfile.mktemp() + '.gz'
with gzip.open(fname, 'wb') as fpt:
for x, y in zip(X, cl):
a = {k: v for k, v in enumerate(x)}
a['klass'] = int(y)
a['num_terms'] = len(x)
try:
fpt.write(bytes(json.dumps(a) + '\n', encoding='utf-8'))
except TypeError:
fpt.write(json.dumps(a) + '\n')
print("termine con el json")
sys.argv = ['EvoDAG', '-C', '-Poutput.evodag', '--json',
'-e1', '-p3', '-r2', fname]
params()
os.unlink(fname)
print(open('output.evodag').read())
os.unlink('output.evodag')
def test_input_type_constraint_C():
from EvoDAG.command_line import params
from EvoDAG.utils import PARAMS
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-C', '--parameters', 'cache.evodag',
'--verbose=0', '-p3', '-e2', '-r2', fname
]
c = params(output=True)
_params = PARAMS.copy()
c.if_type_contraint(_params)
print(_params['input_functions'], 'MultipleVariables')
assert len(_params['input_functions']) == 4
def test_params():
import os
import gzip
from EvoDAG.command_line import params
import json
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '--parameters', 'cache.evodag.gz', '-p3', '-e2', '-r',
'2', fname
]
params()
os.unlink(fname)
assert os.path.isfile('cache.evodag.gz')
with gzip.open('cache.evodag.gz', 'rb') as fpt:
try:
d = fpt.read()
a = json.loads(str(d, encoding='utf-8'))
except TypeError:
a = json.loads(d)
os.unlink('cache.evodag.gz')
assert len(a) == len([x for x in a if 'fitness' in x])
print(a)
default_nargs()
def test_time_limit():
import os
from EvoDAG.command_line import params, train
import json
fname = training_set()
sys.argv = ['EvoDAG', '-C', '--parameters',
'cache.evodag', '-p3', '-e2',
'--time-limit', '10',
'-r', '2', fname]
params()
sys.argv = ['EvoDAG', '--parameters', 'cache.evodag',
'-n2',
'--model', 'model.evodag',
'--test', fname, fname]
train()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'time_limit' in a
os.unlink('cache.evodag')
assert os.path.isfile('model.evodag')
os.unlink('model.evodag')
default_nargs()
def test_time_limit():
import os
from EvoDAG.command_line import params, train
import json
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '--parameters', 'cache.evodag', '-p3', '-e2',
'--time-limit', '10', '-r', '2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--model',
'model.evodag', '--test', fname, fname
]
train()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'time_limit' in a
os.unlink('cache.evodag')
assert os.path.isfile('model.evodag')
os.unlink('model.evodag')
default_nargs()
def test_predict():
import os
from EvoDAG.command_line import params, train, predict
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '--parameters', 'cache.evodag.gz', '-p3', '-e2', '-r',
'2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag.gz', '-n2', '--model',
'model.evodag', '--test', fname, fname
]
train()
sys.argv = [
'EvoDAG', '--output', 'output.evodag', '--model', 'model.evodag', fname
]
predict()
os.unlink(fname)
os.unlink('cache.evodag.gz')
os.unlink('model.evodag')
assert os.path.isfile('output.evodag')
os.unlink('output.evodag')
default_nargs()
def test_predict_cpu():
import os
from EvoDAG.command_line import params, train, predict
fname = training_set()
sys.argv = [
'EvoDAG', '-C', '--parameters', 'cache.evodag', '-p3', '-e1', '-r2',
fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--model',
'model.evodag', '--test', fname, fname
]
train()
sys.argv = [
'EvoDAG', '--output', 'output.evodag', '--decision-function', '-u2',
'--model', 'model.evodag', fname
]
predict()
os.unlink(fname)
os.unlink('cache.evodag')
os.unlink('model.evodag')
os.unlink('output.evodag')
default_nargs()
def test_random_generations():
import os
import json
from EvoDAG.command_line import params
fname = training_set()
sys.argv = [
'EvoDAG',
'-C',
'--parameters',
'cache.evodag',
'-p3',
'-e2',
# '--random-generations', '1',
'-r',
'2',
fname
]
params()
os.unlink(fname)
with open('cache.evodag') as fpt:
a = json.loads(fpt.read())[0]
assert 'random_generations' in a
os.unlink('cache.evodag')
default_nargs()
def test_model_min_size():
import os
from EvoDAG.command_line import params, train
import gzip
import pickle
fname = mo_training_set()
sys.argv = [
'EvoDAG', '--output-dim=3', '-R', '--parameters', 'cache.evodag',
'-p3', '-e2', '-r2', fname
]
params()
sys.argv = [
'EvoDAG', '--parameters', 'cache.evodag', '-n2', '--output-dim=3',
'--min-size=6', '--model', 'model.evodag', '--test', fname, fname
]
train()
os.unlink('cache.evodag')
with gzip.open('model.evodag') as fpt:
m = pickle.load(fpt)
for x in m.models:
print(x.size)
assert x.size >= 6
os.unlink('model.evodag')
default_nargs()
