def gaussian_naive_bayes(self):
"""
for gnb, i train on the training data using different :
1) var_smoothing
:return: test accuracy of the gnb best model
"""
# define parameters
# var_smoothing = np.logspace(start=-9, stop=-3, base=10, num=7, dtype=np.float32)
# best result over all var_smoothing: 0.001
# scale down parameters around its best result (1st round)
np.random.seed(0)
scale = 0.002
loc = 0.001
var_smoothing = loc + scipy.stats.truncnorm.rvs(
-loc / scale, np.infty, size=5,
scale=scale) # To skip negative values
# get the best validated model
gnb = Gaussian_naive_bayes(x_train=self.x_train,
y_train=self.y_train,
cv=5,
var_smoothing=var_smoothing,
grid_search=True)
# print all possible parameter values and the best parameters
# gnb.print_parameter_candidates()
# gnb.print_best_estimator()
# return the accuracy score
return (gnb.evaluate(data=self.x_train, targets=self.y_train),
gnb.evaluate(data=self.x_test, targets=self.y_test))
def gaussian_naive_bayes(self):
"""
for gnb, i train on the training data using different :
1) var_smoothing
:return: ((accuracy_train, recall_train, precision_train),
(accuracy_test, recall_test, precision_test))
"""
# define parameters
# var_smoothing = np.logspace(start=-9, stop=-3, base=10, num=7, dtype=np.float32)
# best result over all var_smoothing: 1e-08
# scale down parameters around its best result
np.random.seed(0)
var_smoothing = norm.rvs(loc=1e-8, scale=1e-8,
size=5).astype(np.float32)
# get the best validated model
gnb = Gaussian_naive_bayes(x_train=self.x_train,
y_train=self.y_train,
cv=5,
var_smoothing=var_smoothing,
grid_search=True)
# print all possible parameter values and the best parameters
# gnb.print_parameter_candidates()
# gnb.print_best_estimator()
# return the accuracy score
return (gnb.evaluate(data=self.x_train, targets=self.y_train),
gnb.evaluate(data=self.x_test, targets=self.y_test))
def gaussian_naive_bayes(self):
"""
for gnb, i train on the training data using different :
1) var_smoothing
:return: test accuracy of the gnb best model
"""
# define arguments given to GridSearchCV
var_smoothing = np.logspace(start=-9,
stop=-6,
base=10,
num=4,
dtype=np.float32)
# get the best validated model
gnb = Gaussian_naive_bayes(x_train=self.x_train,
y_train=self.y_train,
cv=3,
n_jobs=6,
var_smoothing=var_smoothing,
grid_search=True)
# print all possible parameter values and the best parameters
# gnb.print_parameter_candidates()
# gnb.print_best_estimator()
# return the accuracy score
return (gnb.evaluate(data=self.x_train,
targets=self.y_train,
average='macro'),
gnb.evaluate(data=self.x_test,
targets=self.y_test,
average='macro'))
def gaussian_naive_bayes(self):
"""
for gnb, i train on the training data using different :
1) var_smoothing
:return: test accuracy of the gnb best model
"""
# define parameters
# var_smoothing = np.logspace(start=-3, stop=2, base=10, num=6, dtype=np.float32)
# best result over all var_smoothing: 100.0
# scale down parameters around its best result
np.random.seed(0)
var_smoothing = norm.rvs(loc=100.0, scale=20,
size=5).astype(np.float32)
# get the best validated model
gnb = Gaussian_naive_bayes(x_train=self.x_train,
y_train=self.y_train,
cv=5,
var_smoothing=var_smoothing,
grid_search=True)
# print all possible parameter values and the best parameters
gnb.print_parameter_candidates()
gnb.print_best_estimator()
# return the accuracy score
return gnb.accuracy_score(x_test=self.x_test, y_test=self.y_test)
def gaussian_naive_bayes(self):
priors = [(1, ), (20, ), (50, ), (100, )]
gnb = Gaussian_naive_bayes(x_train=self.x_train,
y_train=self.y_train,
cv=3,
n_iter=30,
var_smoothing=np.logspace(start=-9,
stop=-6,
base=10,
num=4,
dtype=np.float32),
grid_search=True)
# print all possible parameter values and the best parameters
#gnb.print_parameter_candidates()
#gnb.print_best_estimator()
# return the accuracy score
return (gnb.evaluate(data=self.x_train, targets=self.y_train),
gnb.evaluate(data=self.x_test, targets=self.y_test))
def gaussian_naive_bayes(self):
var_smoothing = np.logspace(start=-9,
stop=-6,
base=10,
num=4,
dtype=np.float32)
gnb = Gaussian_naive_bayes(x_train=self.x_train,
y_train=self.y_train,
cv=3,
n_jobs=-1,
var_smoothing=var_smoothing,
grid_search=True)
# gnb.print_parameter_candidates()
# gnb.print_best_estimator()
# return the accuracy score
return (gnb.evaluate(data=self.x_train,
targets=self.y_train,
average='micro'),
gnb.evaluate(data=self.x_test,
targets=self.y_test,
average='micro'))