-
Notifications
You must be signed in to change notification settings - Fork 1
/
algorithm.py
253 lines (236 loc) · 10.1 KB
/
algorithm.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
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
import logging
import os
import warnings
from datetime import datetime
from multiprocessing import cpu_count
from threading import Thread
from typing import Any, Dict, List
import pandas as pd
from causallift import CausalLift
from kedro.extras.datasets.pickle.pickle_dataset import PickleDataSet as PickleLocalDataSet
from pandas import DataFrame
from pandas.core.common import SettingWithCopyWarning
from sklearn.exceptions import ConvergenceWarning, UndefinedMetricWarning
from core.confs import path
from core.enums.dataset import OutcomeType
from core.enums.definition import ColumnDefinition
from core.functions.common.etc import random_str
from core.functions.common.file import delete_file
from plugins.common.algorithm import Algorithm
from plugins.common.dataset import get_encoded_dfs_by_activity
# Enable logging
logger = logging.getLogger(__name__)
# Ignore warnings caused by the causallift package itself
warnings.simplefilter("ignore", category=DeprecationWarning)
warnings.simplefilter("ignore", category=ConvergenceWarning)
warnings.simplefilter("ignore", category=SettingWithCopyWarning)
warnings.simplefilter("ignore", category=UndefinedMetricWarning)
os.environ["PYTHONWARNINGS"] = "ignore"
class CausalLiftAlgorithm(Algorithm):
def __init__(self, algo_data: Dict[str, Any]):
super().__init__(algo_data)
self.__training_dfs: Dict[int, DataFrame] = {}
def preprocess(self) -> str:
# Pre-process the data
self.__training_dfs, data = get_encoded_dfs_by_activity(
original_df=self.get_df(),
encoding_type=self.get_parameter_value("encoding"),
outcome_type=OutcomeType.LABELLED,
include_treatment=True,
for_test=False,
existing_data={}
)
for key in data:
if key in {"mapping", "lb"}:
self.set_data_value(key, data[key])
return ""
def train(self) -> str:
# Train the model
self.set_data_value("training_dfs", self.__training_dfs)
return ""
def predict(self, prefix: List[dict]) -> dict:
# Predict the result by using the given prefix
length = len(prefix)
training_df = self.get_data()["training_dfs"].get(length)
if training_df is None:
return self.get_null_output("The model is not trained for the given prefix length")
# Get the test df
raw_test_df = pd.DataFrame(prefix)
test_df = list(get_encoded_dfs_by_activity(
original_df=raw_test_df,
encoding_type=self.get_parameter_value("encoding"),
outcome_type=OutcomeType.LABELLED,
include_treatment=False,
for_test=True,
existing_data=self.get_data()
)[0].values())[0]
# Get the CATE using two models approach
result_df = self.get_result(training_df, test_df)
proba_if_treated = round(result_df["Proba_if_Treated"].values[0].item(), 4)
proba_if_untreated = round(result_df["Proba_if_Untreated"].values[0].item(), 4)
cate = round(result_df["CATE"].values[0].item(), 4)
output = {
"proba_if_treated": proba_if_treated,
"proba_if_untreated": proba_if_untreated,
"cate": cate,
"treatment": self.get_additional_info_value("treatment_definition")
}
return {
"date": datetime.now().isoformat(),
"type": self.get_basic_info()["prescription_type"],
"output": output,
"plugin": {
"name": self.get_basic_info()["name"],
"model": f"{self.get_parameter_value('encoding')}-length-{length}",
}
}
def predict_df(self, df: DataFrame) -> dict:
# Predict the result by using the given dataframe
result = {}
result_dfs: Dict[int, DataFrame] = {}
# Get the test df for each length
test_dfs, _ = get_encoded_dfs_by_activity(
original_df=df,
encoding_type=self.get_parameter_value("encoding"),
outcome_type=OutcomeType.LABELLED,
include_treatment=False,
for_test=True,
existing_data=self.get_data()
)
# Get the result for each length
threads = []
count_of_length = len(test_dfs)
if count_of_length <= 50:
for length, test_df in test_dfs.items():
training_df = self.get_data()["training_dfs"].get(length)
if training_df is None:
continue
t = Thread(target=self.get_result_thread, args=(self, result_dfs, length, training_df, test_df))
t.start()
threads.append(t)
for t in threads:
t.join()
else:
for length, test_df in test_dfs.items():
training_df = self.get_data()["training_dfs"].get(length)
if training_df is None:
continue
result_df = self.get_result(training_df, test_df)
result_dfs[length] = result_df
# Merge the result
if len(result_dfs) <= 0:
return result
for length, result_df in result_dfs.items():
treatment_definition = self.get_additional_info_value("treatment_definition")
prescription_type = self.get_basic_info()["prescription_type"]
plugin_name = self.get_basic_info()["name"]
model_code = f"{self.get_parameter_value('encoding')}-length-{length}"
for _, row in result_df.iterrows():
case_id = self.get_case_id(row)
proba_if_treated = round(row["Proba_if_Treated"].item(), 4)
proba_if_untreated = round(row["Proba_if_Untreated"].item(), 4)
cate = round(row["CATE"].item(), 4)
output = {
"proba_if_treated": proba_if_treated,
"proba_if_untreated": proba_if_untreated,
"cate": cate,
"treatment": treatment_definition
}
result[case_id] = {
"date": datetime.now().isoformat(),
"type": prescription_type,
"output": output,
"plugin": {
"name": plugin_name,
"model": model_code
}
}
return result
@staticmethod
def get_result(training_df: DataFrame, test_df: DataFrame) -> DataFrame:
cols_features = [x for x in training_df.columns
if x not in {ColumnDefinition.OUTCOME, ColumnDefinition.TREATMENT, ColumnDefinition.CASE_ID}]
temp_dir = f"{path.TEMP_PATH}/{random_str(16)}"
try:
n_jobs = cpu_count() if cpu_count() <= 8 else 8
cl = CausalLift(
train_df=training_df,
test_df=test_df,
enable_ipw=True,
logging_config=None,
cols_features=cols_features,
col_treatment=ColumnDefinition.TREATMENT,
col_outcome=ColumnDefinition.OUTCOME, verbose=0,
dataset_catalog=dict(
propensity_model=PickleLocalDataSet(
filepath=f"{temp_dir}/propensity_model.pickle",
version=None
),
uplift_models_dict=PickleLocalDataSet(
filepath=f"{temp_dir}/uplift_models_dict.pickle",
version=None
)
),
uplift_model_params=dict( # type: ignore
search_cv="sklearn.model_selection.GridSearchCV",
estimator="xgboost.XGBClassifier",
scoring=None,
cv=3,
return_train_score=False,
n_jobs=n_jobs,
param_grid=dict(
random_state=[0],
max_depth=[3],
learning_rate=[0.1],
n_estimators=[100],
verbose=[0],
objective=["binary:logistic"],
booster=["gbtree"],
n_jobs=[-1],
nthread=[None],
gamma=[0],
min_child_weight=[1],
max_delta_step=[0],
subsample=[1],
colsample_bytree=[1],
colsample_bylevel=[1],
reg_alpha=[0],
reg_lambda=[1],
scale_pos_weight=[1],
base_score=[0.5],
missing=[None],
),
),
propensity_model_params=dict( # type: ignore
search_cv="sklearn.model_selection.GridSearchCV",
estimator="sklearn.linear_model.LogisticRegression",
scoring=None,
cv=3,
return_train_score=False,
n_jobs=n_jobs,
param_grid=dict(
random_state=[0],
C=[0.1, 1, 10],
class_weight=[None],
dual=[False],
fit_intercept=[True],
intercept_scaling=[1],
max_iter=[100],
multi_class=["ovr"],
n_jobs=[1],
penalty=["l1", "l2"],
solver=["liblinear"],
tol=[0.0001],
warm_start=[False],
),
)
)
_, result_df = cl.estimate_cate_by_2_models()
finally:
delete_file(temp_dir)
return result_df
@staticmethod
def get_result_thread(self, result_dfs: Dict[int, DataFrame], length: int,
training_df: DataFrame, test_df: DataFrame):
# Get the CATE using two models approach
result_dfs[length] = self.get_result(training_df, test_df)