-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathregression_trainer.py
148 lines (130 loc) · 7.02 KB
/
regression_trainer.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
import pickle
import os
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.metrics import explained_variance_score
from sklearn.metrics import mean_squared_error
from sklearn.linear_model import LinearRegression
from sklearn.linear_model import ElasticNet
from sklearn.kernel_ridge import KernelRidge
from sklearn.svm import SVR
from sklearn.neighbors import RadiusNeighborsRegressor
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.neural_network import MLPRegressor
from sklearn.metrics import r2_score
class Regression:
def __init__(self):
print("Training Initialized for Continuous Variable:")
def train(self, data, var, info):
print(var)
train_metrics = dict()
method_list = info[0]
default_active = info[1]
params = info[2]
try:
if not os.path.exists(var):
os.makedirs(var)
for count, method in enumerate(method_list):
reg = getattr(self, "train_" + method)(default_active[count], params[count])
train_metrics[method] = self.train_data(data, var, method, reg)
return train_metrics
except AttributeError:
print("The continuous training method '" + method + "' does not exist, kindly check the config files!")
def train_data(self, data, var, method, reg):
try:
print("Training " + var + " with " + method + " regression method.")
y = np.array(data[var])
x = np.array(data.drop(columns=[var]))
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.33, random_state=42)
reg.fit(x_train, y_train)
y_pred = reg.predict(x_test)
pkl_filename = method + ".pkl"
with open(var + "/" + pkl_filename, 'wb') as file:
pickle.dump(reg, file)
metric_dict = {
"variance_score": explained_variance_score(y_test, y_pred, multioutput='uniform_average'),
"r2_score": r2_score(y_test, y_pred),
"root_mean_sq_err": mean_squared_error(y_test, y_pred, squared=False)
}
print("Training of " + var + " with " + method + " regression method complete!")
return metric_dict
except ValueError:
print("Training of " + var + " with " + method + " regression method Incomplete!")
return {
"variance_score": 0,
"r2_score": 0,
"root_mean_sq_err": 9999
}
def train_linear_reg(self, default, param):
if default:
reg_trainer = LinearRegression()
else:
reg_trainer = LinearRegression(fit_intercept=param["fit_intercept"],
normalize=param["normalize"], positive=param["positive"])
return reg_trainer
def train_elastic_net(self, default, param):
if default:
reg_trainer = ElasticNet(random_state=0)
else:
reg_trainer = ElasticNet(alpha=param["alpha"], l1_ratio=param["l1_ratio"],
fit_intercept=param["fit_intercept"], normalize=param["normalize"],
max_iter=param["max_iter"], tol=param["tol"],
positive=param["positive"], random_state=param["random_state"],
selection=param["selection"])
return reg_trainer
def train_kernel_ridge(self, default, param):
if default:
reg_trainer = KernelRidge(alpha=0.1, kernel='polynomial', degree=2)
else:
reg_trainer = KernelRidge(alpha=param["alpha"], kernel=param["kernel"],
gamma=param["gamma"], degree=param["degree"], coef0=param["coef0"])
return reg_trainer
def train_support_vector(self, default, param):
if default:
reg_trainer = SVR(kernel='rbf')
else:
reg_trainer = SVR(kernel=param["kernel"], degree=param["degree"],
gamma=param["gamma"], coef0=param["coef0"],
tol=param["tol"], C=param["C"], epsilon=param["epsilon"],
shrinking=param["shrinking"], cache_size=param["cache_size"],
verbose=param["verbose"], max_iter=param["max_iter"])
return reg_trainer
def train_radius_neighbor(self, default, param):
if default:
reg_trainer = RadiusNeighborsRegressor(radius=1.0)
else:
reg_trainer = RadiusNeighborsRegressor(radius=param["radius"], weights=param["weights"],
algorithm=param["algorithm"], leaf_size=param["leaf_size"],
p=param["p"], metric=param["metric"], n_jobs=param["n_jobs"])
return reg_trainer
def train_gradient_boost(self, default, param):
if default:
reg_trainer = GradientBoostingRegressor(random_state=0)
else:
reg_trainer = GradientBoostingRegressor(
loss=param["loss"], learning_rate=param["learning_rate"],
n_estimators=param["n_estimators"], subsample=param["subsample"],
criterion=param["criterion"], min_samples_split=param["min_samples_split"],
min_samples_leaf=param["min_samples_leaf"], min_weight_fraction_leaf=param["min_weight_fraction_leaf"],
max_depth=param["max_depth"], min_impurity_decrease=param["min_impurity_decrease"],
random_state=param["random_state"], max_features=param["max_features"],
alpha=param["alpha"], verbose=param["verbose"], max_leaf_nodes=param["max_leaf_nodes"],
validation_fraction=param["validation_fraction"], n_iter_no_change=param["n_iter_no_change"],
tol=param["tol"], ccp_alpha=param["ccp_alpha"])
return reg_trainer
def train_neural_MLP(self, default, param):
if default:
reg_trainer = MLPRegressor(random_state=0, max_iter=500)
else:
reg_trainer = MLPRegressor(
hidden_layer_sizes=tuple(param["hidden_layer_sizes"]), activation=param["activation"],
solver=param["solver"], alpha=param["alpha"], batch_size=param["batch_size"],
learning_rate=param["learning_rate"], learning_rate_init=param["learning_rate_init"],
power_t=param["power_t"], max_iter=param["max_iter"], shuffle=param["shuffle"],
random_state=param["random_state"], tol=param["tol"], verbose=param["verbose"],
warm_start=param["warm_start"], momentum=param["momentum"],
nesterovs_momentum=param["nesterovs_momentum"], early_stopping=param["early_stopping"],
validation_fraction=param["validation_fraction"], beta_1=param["beta_1"],
beta_2=param["beta_2"], epsilon=param["epsilon"], n_iter_no_change=param["n_iter_no_change"],
max_fun=param["max_fun"])
return reg_trainer