/
prediction.py
181 lines (146 loc) · 7.1 KB
/
prediction.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
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
import inspect
import os
import pickle
from types import FunctionType
import numpy as np
import pandas as pd
from sklearn.cross_validation import KFold
from dataset import Dataset
from feature_selection import *
from target_transform import *
from utils import md5, touch, normalized_gini
class Prediction(object):
'''Prediction object for Liberty Insurance Competition.
'''
def __init__(self, dataset, model, base_path, target_transform=BaseTargetTransform,
feature_selector=BaseFeatureSelector, save=True):
self.dataset = dataset
self.model = model
self.target_transform = target_transform
self.feature_selector = feature_selector()
self.save = save
self.save_path = self._generate_filename(base_path)
def _load_data(self):
X_train, y_train, X_test = self.dataset.generate()
return X_train, y_train, X_test
def _extract_model_parameters(self):
if hasattr(self.model, 'estimator_params'):
return dict((param, getattr(self.model, param)) for param in
self.model.estimator_params)
elif hasattr(self.model, '_get_param_names'):
return dict((param, getattr(self.model, param)) for param in
self.model._get_param_names())
else:
raise NotImplemented, 'model must implement estimator_params' \
' or _get_param_names()'
def _get_model_name(self):
return self.model.__class__.__name__
def _get_dataset_func_name(self):
return self.dataset.func.__name__
def _get_feature_selector_name(self):
return self.feature_selector.__class__.__name__
def _get_target_transform_name(self):
return self.target_transform.__name__
@staticmethod
def _get_init_args_from_class(my_class):
init_args = inspect.getargspec(my_class.__init__).args[1:]
init_args_dict = {a: my_class.__dict__[a] for a in init_args}
return init_args_dict
def _hash_model_params(self):
params = self._extract_model_parameters()
params_str = ' '.join(['%s=%s' % (k, v) for k, v in params.items()])
return md5(params_str)
def _hash_feature_selector_params(self):
params = repr(self._get_init_args_from_class(self.feature_selector))
return md5(params)
def _hash_function_kwargs(self):
ret = []
for k, v in self.dataset.kwargs.iteritems():
if type(v) == FunctionType:
ret.append(str(k) + '_' + v.__name__)
elif hasattr(v, '__name__'):
ret.append(str(k) + '_' + v.__name__ + '_' + str(v))
elif hasattr(v.__class__, '__name__'):
ret.append(str(k) + '_' + v.__class__.__name__ + '_' + str(v))
else:
ret.append(str(k) + '_' + str(v))
return md5(''.join(ret))
def _generate_filename(self, base_path):
filename_args = {
'dataset_func': self._get_dataset_func_name(),
'hashed_func_kwargs': self._hash_function_kwargs(),
'model_name': self._get_model_name(),
'hashed_model_params': self._hash_model_params(),
'target_transform_name': self._get_target_transform_name(),
'feature_selector_name': self._get_feature_selector_name(),
'hashed_feature_selector_params': self._hash_feature_selector_params()
}
filename = '%(dataset_func)s_%(hashed_func_kwargs)s_%(model_name)s_%(hashed_model_params)s_%(target_transform_name)s_%(feature_selector_name)s_%(hashed_feature_selector_params)s.pkl' % filename_args
return os.path.join(base_path, filename)
def _generate_oof_predictions(self, X_train, y_train, feature_selection_func=None, **kwargs):
train_n = X_train.shape[0]
cv = KFold(n=train_n, n_folds=10, random_state=123)
cv_scores = []
oof_predictions = np.zeros(shape=(train_n,))
for fold, (tr_idx, te_idx) in enumerate(cv):
X_train_ = X_train.iloc[tr_idx]
y_train_ = self.target_transform.transform(y_train.iloc[tr_idx])
X_test_ = X_train.iloc[te_idx]
y_test_ = y_train.iloc[te_idx]
# select features
self.feature_selector.fit(X_train_, y_train_)
X_train_ = self.feature_selector.transform(X_train_)
X_test_ = self.feature_selector.transform(X_test_)
self.model.fit(X_train_, y_train_)
preds_k = self.target_transform.transform_back(self.model.predict(X_test_))
oof_predictions[te_idx] = preds_k
gini_k = normalized_gini(y_test_, preds_k)
cv_scores.append(gini_k)
print 'Fold %d: %.4f' % (fold + 1, gini_k)
oof_gini = normalized_gini(y_train, oof_predictions)
print 'Final: %.4f' % (oof_gini)
return oof_predictions, oof_gini, cv_scores
def _generate_lb_predictions(self, X_train, y_train, X_test):
train_n, test_n = X_train.shape[0], X_test.shape[0]
index = ['train']*train_n + ['test']*test_n
predictions = pd.DataFrame(np.zeros(shape=(train_n + test_n,)), index=index)
# select features
final_features = self.feature_selector.return_final_features()
X_train, X_test = X_train[final_features], X_test[final_features]
self.model.fit(X_train, self.target_transform.transform(y_train))
lb_predictions = self.target_transform.transform_back(self.model.predict(X_test))
return lb_predictions
def cross_validate(self, feature_selection_func=None, **kwargs):
if self.save is True:
if os.path.exists(self.save_path):
try:
pred = pickle.load(open(self.save_path, 'rb'))
oof_gini = pred['normalized_gini']
print '`{}` exists. CV: {}'.format(
self.save_path.split('/')[-1],
np.round(oof_gini, 4)
)
return
except EOFError:
pass
# reserve the filename
touch(self.save_path)
X_train, y_train, X_test = self._load_data()
oof_predictions, oof_gini, cv_scores = self._generate_oof_predictions(X_train, y_train)
lb_predictions = self._generate_lb_predictions(X_train, y_train, X_test)
# save
if self.save is True:
to_save = {
'oof_predictions': oof_predictions,
'lb_predictions': lb_predictions,
'normalized_gini': oof_gini,
'normalized_gini_cv': cv_scores,
'model_params': self._extract_model_parameters(),
'model_name': self._get_model_name(),
'dataset_func': self._get_dataset_func_name(),
'dataset_params': self.dataset.kwargs,
'target_transform': self._get_target_transform_name(),
'feature_selector': self._get_feature_selector_name(),
'feature_selector_params': self._get_init_args_from_class(self.feature_selector)
}
pickle.dump(to_save, open(self.save_path, 'wb'))