def run(dataset_train, dataset_test, pop_size, gens, cross_rate, fb, max_time=1200):
Xtrain, ytrain = dataset_train[:, :-1], dataset_train[:, -1]
Xtest, ytest = dataset_test[:, :-1], dataset_test[:, -1]
est_gp = Feat(obj="fitness,complexity",
pop_size=pop_size,
gens=gens,
max_time=max_time,
max_stall=50,
batch_size=10000,
ml = "LinearRidgeRegression",
sel='lexicase',
surv='nsga2',
max_depth=10,
max_dim=min([Xtrain.shape[1]*2,50]),
#random_state=random_seed,
functions="+,-,*,/,sqrt,sin,cos,tanh,exp,log,^,x,kd",
otype="f",
backprop=True,
iters=10,
n_threads=1,
verbosity=1,
# tuned parameters
cross_rate= cross_rate,
fb = fb,
root_xo_rate = 0.75,
softmax_norm = False
)
est_gp.fit(Xtrain, ytrain)
return RMSE(est_gp.predict(Xtrain), ytrain), RMSE(est_gp.predict(Xtest), ytest), est_gp.get_model()
def test_saving_loading(self):
self.debug("Pickle Feat object")
reg = clone(self.reg)
reg.fit(self.X, self.yr)
initial_pred = reg.predict(self.X)
reg.save('Feat_tmp.json')
loaded_reg = Feat().load('Feat_tmp.json')
# print('loaded_reg:',type(loaded_reg).__name__)
loaded_pred = loaded_reg.predict(self.X)
# print('initial pred:',initial_pred)
# print('loaded pred:',loaded_pred)
diff = np.abs(initial_pred-loaded_pred)
for i,d in enumerate(diff):
if d > 0.0001:
print('pred:',initial_pred[i],'loaded:',loaded_pred[i],
'diff:',d)
assert(d < 0.0001)
# assert(all([ip==lp for ip,lp in zip(initial_pred, loaded_pred)]))
assert(reg.get_representation() == loaded_reg.get_representation())
assert(reg.get_model() == loaded_reg.get_model())
assert((reg.get_coefs() == loaded_reg.get_coefs()).all())
loaded_params = loaded_reg.get_params()
# print('\n',10*'=','\n')
# print('loaded_params:')
# for k,v in loaded_params.items():
# print(k,':',v)
for k,v in reg.get_params().items():
if k not in loaded_params.keys():
print(k,'not in ',loaded_params.keys())
assert(k in loaded_params.keys())
if isinstance(v,float):
if np.abs(loaded_params[k] - v) > 0.0001:
print('loaded_params[',k,'] =',
loaded_params[k], '\nwhich is different from:', v)
assert(np.abs(loaded_params[k] - v) < 0.0001)
elif loaded_params[k] != v:
print('loaded_params[',k,'] =',
loaded_params[k], '\nwhich is different from:', v)
assert(loaded_params[k] == v)
loaded_reg.fit(self.X, self.yr)
class TestFeatWrapper(unittest.TestCase):
def setUp(self):
self.v = verbosity
self.clf = Feat(verbosity=verbosity, n_threads=1)
diabetes = load_diabetes()
self.X = diabetes.data
self.y = diabetes.target
#Test 1: Assert the length of labels returned from predict
def test_predict_length(self):
self.debug("Fit the Data")
self.clf.fit(self.X, self.y)
self.debug("Predicting the Results")
pred = self.clf.predict(self.X)
self.debug("Comparing the Length of labls in Predicted vs Actual ")
expected_length = len(self.y)
actual_length = len(pred)
self.assertEqual(actual_length, expected_length)
#Test 2: Assert the length of labels returned from fit_predict
def test_fitpredict_length(self):
self.debug("Calling fit_predict from Feat")
pred = self.clf.fit_predict(self.X, self.y)
self.debug("Comparing the length of labls in fit_predict vs actual ")
expected_length = len(self.y)
actual_length = len(pred)
self.assertEqual(actual_length, expected_length)
#Test 3: Assert the length of labels returned from transform
def test_transform_length(self):
self.debug("Calling fit")
self.clf.fit(self.X, self.y)
trans_X = self.clf.transform(self.X)
self.debug(
"Comparing the length of labls in transform vs actual feature set "
)
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual(actual_value, expected_value)
#Test 4: Assert the length of labels returned from fit_transform
def test_fit_transform_length(self):
self.debug("In wrappertest.py...Calling fit transform")
trans_X = self.clf.fit_transform(self.X, self.y)
self.debug(
"Comparing the length of labls in transform vs actual feature set "
)
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual(actual_value, expected_value)
#Test 5: Transform with Z
def test_transform_length_z(self, zfile=None, zids=None):
self.debug("Calling fit")
self.clf.fit(self.X, self.y)
trans_X = self.clf.transform(self.X, zfile, zids)
self.debug(
"Comparing the length of labls in transform vs actual feature set "
)
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual(actual_value, expected_value)
def debug(self, message):
if (self.v > 0):
print(message)
def test_coefs(self):
self.debug("In wrappertest.py...Calling test_coefs")
self.clf.fit(self.X, self.y)
coefs = self.clf.get_coefs()
self.assertTrue(len(coefs) > 0)
def test_dataframe(self):
self.debug("In wrappertest.py...Calling test_dataframe")
dfX = pd.DataFrame(
data=self.X,
columns=['fishy' + str(i) for i in np.arange(self.X.shape[1])],
index=None)
dfy = pd.DataFrame(data={'label': self.y})
self.clf.fit(dfX, dfy['label'])
assert (self.clf.feature_names == ','.join(dfX.columns).encode())
#Test: Assert the length of labels returned from predict
def test_predict_stats_length(self):
self.debug("Fit the Data")
self.clf.fit(self.X, self.y)
for key in self.clf.stats:
self.assertEqual(len(self.clf.stats[key]), self.clf.gens)
#Test ability to pickle feat model
def test_pickling(self):
self.debug("Pickle Feat object")
with open('test_pickle.pkl', 'wb') as f:
pickle.dump(self.clf, f)
with open('test_pickle.pkl', 'rb') as f:
loaded_clf = pickle.load(f)
assert (loaded_clf.get_params() == self.clf.get_params())
def test_archive(self):
"""test archiving ability"""
self.debug("Test archive")
self.clf.classification = True
self.clf.ml = b'LR'
self.clf.fit(self.X, np.array(self.y > np.median(self.y),
dtype=np.int))
archive = self.clf.get_archive()
preds = self.clf.predict_archive(self.X)
probs = self.clf.predict_proba_archive(self.X)
for arch, pred, prob in zip(archive, preds, probs):
self.assertTrue(arch['id'] == pred['id'])
self.assertTrue(arch['id'] == prob['id'])
def test_lr_l1(self):
"""testing l1 penalized LR"""
self.clf.classification = True
self.clf.ml = b'L1_LR'
self.clf.fit(self.X, np.array(self.y > np.median(self.y),
dtype=np.int))
self.assertEqual(len(self.clf.predict(self.X)), len(self.y))
pop_size=200,
# ml='CART',
ml='LR',
verbosity=1,
shuffle=True,
classification=True,
backprop=True,
random_state=42)
lr = LR()
rocs = []
aucs = []
lr_rocs = []
lr_aucs = []
for train_idx, test_idx in kf.split(X):
clf.fit(X[train_idx], y[train_idx])
lr.fit(X[train_idx], y[train_idx])
probabilities = clf.predict_proba(X[test_idx])
lr_probabilities = lr.predict_proba(X[test_idx])
fpr, tpr, _ = roc_curve(y[test_idx], probabilities[:, 1])
lr_fpr, lr_tpr, _ = roc_curve(y[test_idx], lr_probabilities[:, 1])
aucs.append(auc(fpr, tpr))
lr_aucs.append(auc(lr_fpr, lr_tpr))
rocs.append((fpr, tpr))
lr_rocs.append((lr_fpr, lr_tpr))
import matplotlib.pyplot as plt
import numpy as np
from feat import Feat
from sklearn.model_selection import KFold
df = pd.read_csv('d_example_patients.csv')
df.drop('id', axis=1, inplace=True)
X = df.drop('class', axis=1).values
y = df['class'].values
zfile = 'd_example_patients_long.csv'
kf = KFold(n_splits=3)
kf.get_n_splits(X)
clf = Feat(
max_depth=5,
max_dim=min(50, 2 * X.shape[1]),
verbosity=1,
shuffle=True,
ml='LR',
classification=True,
functions=
"max,+,-,*,/,exp,log,and,or,not,=,<,>,ite,mean,median,min,variance,skew,kurtosis,slope,count",
random_state=42)
scores = []
for train_idx, test_idx in kf.split(X):
clf.fit(X[train_idx], y[train_idx], zfile, train_idx)
scores.append(clf.score(X[test_idx], y[test_idx], zfile, test_idx))
print('scores:', scores)
class TestFeatWrapper(unittest.TestCase):
def setUp(self):
self.v = verbosity
self.clf = Feat(verbosity=self.v)
diabetes = load_diabetes()
self.X = diabetes.data
self.y = diabetes.target
#Test 1: Assert the length of labels returned from predict
def test_predict_length(self):
self.debug("Fit the Data")
self.clf.fit(self.X,self.y)
self.debug("Predicting the Results")
pred = self.clf.predict(self.X)
self.debug("Comparing the Length of labls in Predicted vs Actual ")
expected_length = len(self.y)
actual_length = len(pred)
self.assertEqual( actual_length , expected_length )
#Test 2: Assert the length of labels returned from fit_predict
def test_fitpredict_length(self):
self.debug("Calling fit_predict from Feat")
pred = self.clf.fit_predict(self.X,self.y)
self.debug("Comparing the length of labls in fit_predict vs actual ")
expected_length = len(self.y)
actual_length = len(pred)
self.assertEqual( actual_length , expected_length )
#Test 3: Assert the length of labels returned from transform
def test_transform_length(self):
self.debug("Calling fit")
self.clf.fit(self.X,self.y)
trans_X = self.clf.transform(self.X)
self.debug("Comparing the length of labls in transform vs actual feature set ")
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual( actual_value , expected_value )
#Test 4: Assert the length of labels returned from fit_transform
def test_fit_transform_length(self):
self.debug("In wrappertest.py...Calling fit transform")
trans_X = self.clf.fit_transform(self.X,self.y)
self.debug("Comparing the length of labls in transform vs actual feature set ")
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual( actual_value , expected_value )
#Test 5: Transform with Z
def test_transform_length_z(self,zfile=None,zids=None):
self.debug("Calling fit")
self.clf.fit(self.X,self.y)
trans_X = self.clf.transform(self.X,zfile,zids)
self.debug("Comparing the length of labls in transform vs actual feature set ")
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual( actual_value , expected_value )
def debug(self,message):
if ( self.v > 0 ):
print (message)
def test_coefs(self):
self.debug("In wrappertest.py...Calling test_coefs")
self.clf.fit(self.X,self.y)
coefs = self.clf.get_coefs()
print('coefs:',coefs)
self.assertTrue( len(coefs)>0 )
import pandas as pd
from pmlb import fetch_data
df = pd.read_csv('mnist.csv', sep='\t')
print(df.columns)
X = df.drop('class', axis=1).values
y = df['class'].values
from feat import Feat
ft = Feat(classification=True, verbosity=2)
ft.fit(X[:60000], y[:60000])
print(ft.score(X[60000:], y[60000:]))
pop_size=100,
verbosity=1,
shuffle=True,
ml='LR',
classification=True,
feature_names=','.join(df.drop('class', axis=1).columns),
functions="+,-,*,/,exp,log,and,or,not,=,<,<=,>,>=,ite,split,split_c,"
"mean,median,max,min,variance,skew,kurtosis,slope,count",
backprop=True,
iters=10,
random_state=42)
scores = []
for train_idx, test_idx in kf.split(X, y):
# print('train_idx:',train_idx)
clf.fit(X[train_idx], y[train_idx], zfile, train_idx)
scores.append(clf.score(X[test_idx], y[test_idx], zfile, test_idx))
print('scores:', scores)
###################################################################################################
# fit to all data
###################################################################################################
print('fitting longer to all data...')
clf.gens = 20
clf.verbosity = 2
clf.fit(X, y, zfile, np.arange(len(X)))
print('model:', clf.get_model())
##################################################################################################
import pandas as pd
import numpy as np
from feat import Feat
import sys
seed = sys.argv[1]
df = pd.read_csv('../examples/d_heart.csv', sep=',')
df.describe()
X = df.drop('class', axis=1).values
y = df['class'].values
clf = Feat(max_depth=3,
max_dim=1,
gens=100,
pop_size=200,
verbosity=2,
shuffle=True,
classification=True,
functions="+,-,*,/,exp,log,and,or,not,=,<,>,ite",
random_state=seed,
softmax_norm=True)
clf.fit(X, y)
class TestFeatWrapper(unittest.TestCase):
def setUp(self):
self.v = verbosity
self.clf = Feat(verbosity=verbosity, n_threads=1)
diabetes = load_diabetes()
self.X = diabetes.data
self.y = diabetes.target
#Test 1: Assert the length of labels returned from predict
def test_predict_length(self):
self.debug("Fit the Data")
self.clf.fit(self.X,self.y)
self.debug("Predicting the Results")
pred = self.clf.predict(self.X)
self.debug("Comparing the Length of labls in Predicted vs Actual ")
expected_length = len(self.y)
actual_length = len(pred)
self.assertEqual( actual_length , expected_length )
#Test 2: Assert the length of labels returned from fit_predict
def test_fitpredict_length(self):
self.debug("Calling fit_predict from Feat")
pred = self.clf.fit_predict(self.X,self.y)
self.debug("Comparing the length of labls in fit_predict vs actual ")
expected_length = len(self.y)
actual_length = len(pred)
self.assertEqual( actual_length , expected_length )
#Test 3: Assert the length of labels returned from transform
def test_transform_length(self):
self.debug("Calling fit")
self.clf.fit(self.X,self.y)
trans_X = self.clf.transform(self.X)
self.debug("Comparing the length of labls in transform vs actual feature set ")
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual( actual_value , expected_value )
#Test 4: Assert the length of labels returned from fit_transform
def test_fit_transform_length(self):
self.debug("In wrappertest.py...Calling fit transform")
trans_X = self.clf.fit_transform(self.X,self.y)
self.debug("Comparing the length of labls in transform vs actual feature set ")
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual( actual_value , expected_value )
#Test 5: Transform with Z
def test_transform_length_z(self,zfile=None,zids=None):
self.debug("Calling fit")
self.clf.fit(self.X,self.y)
trans_X = self.clf.transform(self.X,zfile,zids)
self.debug("Comparing the length of labls in transform vs actual feature set ")
expected_value = self.X.shape[0]
actual_value = trans_X.shape[0]
self.assertEqual( actual_value , expected_value )
def debug(self,message):
if ( self.v > 0 ):
print (message)
def test_coefs(self):
self.debug("In wrappertest.py...Calling test_coefs")
self.clf.fit(self.X,self.y)
coefs = self.clf.get_coefs()
print('coefs:',coefs)
self.assertTrue( len(coefs)>0 )
def test_dataframe(self):
self.debug("In wrappertest.py...Calling test_dataframe")
dfX = pd.DataFrame(data=self.X,columns=['fishy'+str(i)
for i in np.arange(self.X.shape[1])],
index=None)
# print(dfX.head())
# print('dfX.columns:',dfX.columns)
dfy = pd.DataFrame(data={'label':self.y})
self.clf.fit(dfX,dfy['label'])
# print('clf feature_names:',self.clf.feature_names)
# print('dfX.columns:',','.join(dfX.columns).encode())
assert(self.clf.feature_names == ','.join(dfX.columns).encode())
#Test: Assert the length of labels returned from predict
def test_predict_stats_length(self):
self.debug("Fit the Data")
self.clf.fit(self.X,self.y)
print("Num generations is ", self.clf.gens)
for key in self.clf.stats:
print("Length for ", key, "is ", len(self.clf.stats[key]))
self.assertEqual(len(self.clf.stats[key]), self.clf.gens)