def __init__(self, parameters, target, repetitions, db_name, transform=dont_transform):
print("v. 0.3")
fixed_parameters, lower_limits, upper_limits, dtypes = split(parameters)
SimulationSet.setup(dtypes, target, transform)
InputSet.setup(lower_limits, upper_limits, dtypes, fixed_parameters)
self.repetitions = repetitions
self.remote_saudifirms = RemoteSaudiFirms(task=5007, result=5008, kill=5009)
self.local_saudifirms = RemoteSaudiFirms(task=5567, result=5568, kill=5569)
self._db = dataset.connect("sqlite:///%s.sqlite3" % db_name)
parameters = json.dumps(sorted(fixed_parameters.items()))
hash = sha224(parameters).hexdigest()[:7]
overview = self._db['overview']
overview.insert({'hash': hash, 'json_string': parameters})
self.db = self._db[hash]
self.db_name = db_name
def insert(self, input, metric):
"""
inserts an array into the data
:param input: ndarray
:
"""
assert isinstance(input, np.ndarray)
input = InputSet.respect_bounds(input)
if not self.is_in_data(input):
self.inputs.append(input)
self.metrics.append(metric)
def initialize(self, initial_runs):
self.simulations = SimulationSet()
for row in self.db:
try:
input = json.loads(row['input'])
output = json.loads(row['output'])
except TypeError:
logger.info(row['output'])
self.simulations.insert(input, output)
logger.info("loaded %i" % len(self.simulations))
if len(self.simulations) < initial_runs:
middle_point = InputSet.middle_point()
simulation_candidates = centered_latin_hypercube_I(middle_point, np.ones_like(middle_point), InputSet.lb, InputSet.ub, initial_runs)
self.run_add(simulation_candidates)
print("done loading/initializing")
def print_dict(self):
inputset = InputSet()
inputset.insert(InputSet.middle_point())
logger.info(dict(inputset[0]))
def run_input(self, weight=0.5):
self.run_local(InputSet.weight_point(weight))
x_min = float(target[j])
else:
if float(output[j]) < float(target[j]):
if float(output[j]) < x_min:
x_min = float(output[j])
else:
if float(target[j]) < x_min:
x_min = float(target[j])
sum_error += error
mse = sum_error / test.size()
rms = numpy.sqrt(mse)
return rms/(x_max-x_min)
if __name__ == "__main__":
data_set = InputSet()
data_set.load("../data/servo.data")
#<<ID: 487; SCORE: 0.20303 [{q:0.937668}{h:17}{c:144,54,41,81,85,52,140,67,48,138,95,110,6,32,5,86,22}{gamma:0.996954,2.48103,0.602891,3,1.42868,0.0979053,3,1.21504,0.922433,0.357251,0.131358,1.23423,0,0,0.299883,0.263622,1.9044}{lambda:0.0446531}]>>
#centers = [144,54,41,81,85,52,140,68,48,138,95,110,6,32,5,86,22]
#gammas = [0.996954,2.48103,1.602891,3,1.42868,0.0979053,3,1.21504,0.922433,0.357251,0.131358,1.23423,0,0,0.299883,0.263622,1.9044]
#q = 0.937668
#l = 0.0446531
#hnn3 = HNN3(data_set, centers, gammas, q, l, "cls")
#print str(hnn3.get_accuracy()*100) + "%"
#<<ID: 28724; SCORE: 0.180124 [{q:0.254867}{h:41}{c:33,104,136,22,30,83,0,138,12,63,130,17,10,147,18,42,139,152,71,31,45,140,124,109,59,13,20,27,102,1,68,97,132,129,149,34,85,123,87,116,134}{gamma:0.963652,1.22752,0.502414,2.06288,0.957675,0.527096,2.29231,2.58408,0.419801,1.09144,0.742667,0.613447,0.301913,1.21125,0.695696,2.16505,0.0783844,0.550067,1.94488,0.679182,1.83098,0.42336,2.02004,0.700278,0.938683,1.02612,2.87757,0.962745,1.6509,0.88908,2.14539,0.078447,0.064547,1.38313,0.0152284,1.5391,0.87585,0.50174,2.01498,0.384209,1.39222}{lambda:0.140671}]>>
#centers2 = [33,104,136,22,30,83,0,138,12,63,130,17,10,147,18,42,139,152,71,31,45,140,124,109,59,13,20,27,102,1,68,97,132,129,149,34,85,123,87,116,134]
#gammas2 = [0.963652,1.22752,0.502414,2.06288,0.957675,0.527096,2.29231,2.58408,0.419801,1.09144,0.742667,0.613447,0.301913,1.21125,0.695696,2.16505,0.0783844,0.550067,1.94488,0.679182,1.83098,0.42336,2.02004,0.700278,0.938683,1.02612,2.87757,0.962745,1.6509,0.88908,2.14539,0.078447,0.064547,1.38313,0.0152284,1.5391,0.87585,0.50174,2.01498,0.384209,1.39222]
#q2 = 0.254867
#hnn3_2 = HNN3(data_set, centers2, gammas2, q2, "cls")
l = self.genoma[1].genes[0]
centers = self.genoma[3].genes
gammas = self.genoma[4].genes
hnn3 = HNN3(self.data_set, centers, gammas, q, l)
return hnn3
def update_score(self):
if self.data_set.mode == "cls":
self.score = self.__fitness_function_classification()
else:
self.score = self.__fitness_function_regression()
def __str__(self):
s = "<< ID: " + str(self.id) + ", SCORE: " + str(self.score)
for i, v in enumerate(self.genoma):
s = s + " { " + str(v) + " } "
s += ">>"
return s
if __name__ == "__main__":
data_set = InputSet()
data_set.load("hepatitis.data")
ind = IndividualHNN3(data_set)
ind.update_score()
print ind
for i in range(20):
o = ind.mutate()
print o
o.update_score()
def centered_latin_hypercube_I(center, schlinge, lb, ub, samples):
ret_lh = _centered_latin_hypercube(center, schlinge, lb, ub, samples)
candidates = InputSet()
for input in ret_lh:
candidates.insert(input)
return candidates
else:
if float(output[j]) < float(target[j]):
if float(output[j]) < x_min:
x_min = float(output[j])
else:
if float(target[j]) < x_min:
x_min = float(target[j])
sum_error += error
mse = sum_error / test.size()
rms = numpy.sqrt(mse)
return rms / (x_max - x_min)
if __name__ == "__main__":
data_set = InputSet()
data_set.load("../data/servo.data")
#<<ID: 487; SCORE: 0.20303 [{q:0.937668}{h:17}{c:144,54,41,81,85,52,140,67,48,138,95,110,6,32,5,86,22}{gamma:0.996954,2.48103,0.602891,3,1.42868,0.0979053,3,1.21504,0.922433,0.357251,0.131358,1.23423,0,0,0.299883,0.263622,1.9044}{lambda:0.0446531}]>>
#centers = [144,54,41,81,85,52,140,68,48,138,95,110,6,32,5,86,22]
#gammas = [0.996954,2.48103,1.602891,3,1.42868,0.0979053,3,1.21504,0.922433,0.357251,0.131358,1.23423,0,0,0.299883,0.263622,1.9044]
#q = 0.937668
#l = 0.0446531
#hnn3 = HNN3(data_set, centers, gammas, q, l, "cls")
#print str(hnn3.get_accuracy()*100) + "%"
#<<ID: 28724; SCORE: 0.180124 [{q:0.254867}{h:41}{c:33,104,136,22,30,83,0,138,12,63,130,17,10,147,18,42,139,152,71,31,45,140,124,109,59,13,20,27,102,1,68,97,132,129,149,34,85,123,87,116,134}{gamma:0.963652,1.22752,0.502414,2.06288,0.957675,0.527096,2.29231,2.58408,0.419801,1.09144,0.742667,0.613447,0.301913,1.21125,0.695696,2.16505,0.0783844,0.550067,1.94488,0.679182,1.83098,0.42336,2.02004,0.700278,0.938683,1.02612,2.87757,0.962745,1.6509,0.88908,2.14539,0.078447,0.064547,1.38313,0.0152284,1.5391,0.87585,0.50174,2.01498,0.384209,1.39222}{lambda:0.140671}]>>
#centers2 = [33,104,136,22,30,83,0,138,12,63,130,17,10,147,18,42,139,152,71,31,45,140,124,109,59,13,20,27,102,1,68,97,132,129,149,34,85,123,87,116,134]
#gammas2 = [0.963652,1.22752,0.502414,2.06288,0.957675,0.527096,2.29231,2.58408,0.419801,1.09144,0.742667,0.613447,0.301913,1.21125,0.695696,2.16505,0.0783844,0.550067,1.94488,0.679182,1.83098,0.42336,2.02004,0.700278,0.938683,1.02612,2.87757,0.962745,1.6509,0.88908,2.14539,0.078447,0.064547,1.38313,0.0152284,1.5391,0.87585,0.50174,2.01498,0.384209,1.39222]
#q2 = 0.254867
#hnn3_2 = HNN3(data_set, centers2, gammas2, q2, "cls")
q = self.genoma[0].genes[0]
l = self.genoma[1].genes[0]
centers = self.genoma[3].genes
gammas = self.genoma[4].genes
hnn3 = HNN3(self.data_set, centers, gammas, q, l)
return hnn3
def update_score(self):
if self.data_set.mode == "cls":
self.score = self.__fitness_function_classification()
else:
self.score = self.__fitness_function_regression()
def __str__(self):
s = "<< ID: " + str(self.id) + ", SCORE: " + str(self.score)
for i,v in enumerate(self.genoma):
s = s + " { " + str(v) + " } "
s += ">>"
return s
if __name__ == "__main__":
data_set = InputSet()
data_set.load("hepatitis.data")
ind = IndividualHNN3(data_set)
ind.update_score()
print ind
for i in range(20):
o = ind.mutate()
print o
o.update_score()
