-
Notifications
You must be signed in to change notification settings - Fork 0
/
start-workflow.py
351 lines (309 loc) · 15.1 KB
/
start-workflow.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
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
"""
Copyright [2022-23] [Intel Corporation]
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
"""
#!/usr/bin/env python
# coding: utf-8
import argparse
import os, time, gc, sys, glob
import pandas as pd
import numpy as np
import yaml
import time
from src.utils.data_utils import *
very_start = time.time()
PATH_HOME = os.path.dirname(os.path.realpath(__file__))
class WFProcessor:
def __init__(self, file_name, mode):
if mode == 1:
self.log_path = '/workspace/tmp/logs'
self.data_path = '/workspace/data'
self.tmp_path = '/workspace/tmp'
self.config_path = '/workspace/configs'
with open(os.path.join(self.config_path,os.path.basename(file_name)),'r') as file:
config = yaml.safe_load(file)
else:
with open(file_name,'r') as file:
config = yaml.safe_load(file)
self.data_path = config['env']['data_path']
self.tmp_path = os.path.join(config['env']['tmp_path'], 'wf-tmp')
self.log_path = os.path.join(self.tmp_path, 'logs')
self.config_path = config['env']['config_path']
self.num_node = config['env']['num_node']
self.is_multi_nodes = True if self.num_node > 1 else False
self.worker_ips = config['env']['node_ips'][1:]
try:
self.raw_data_path = os.path.join(self.data_path, config['data_preprocess']['input_data_path'])
self.raw_data_format = config['data_preprocess']['input_data_format']
dp_config_file = os.path.join(self.config_path, config['data_preprocess']['dp_config_file'])
self.dp_framework = config['data_preprocess']['dp_framework']
self.processed_data_path = os.path.join(self.data_path, config['data_preprocess']['output_data_path'])
self.processed_data_format = config['data_preprocess']['output_data_format']
self.read_data_processing_steps(dp_config_file)
self.identify_dp_engine()
self.has_dp = True
except Exception as e:
print('Failed to read data preprocessing steps. This is either due to wrong parameters defined in the config file as shown: '+ str(e)
+ ' or there is no need for data preprocessing.')
self.has_dp = False
try:
self.train_data_path = os.path.join(self.data_path, config['training']['input_data_path'])
self.train_data_format = config['training']['input_data_format']
train_config_file = os.path.join(self.config_path, config['training']['train_config_file'])
self.train_framework = config['training']['train_framework']
self.test_backend = config['training']['test_backend']
self.read_training_configs(train_config_file)
try:
self.ray_params = config['training']['ray_params']
except:
self.ray_params = None
self.has_training = True
except Exception as e:
print('Failed to read model training configurations. This is either due to wrong parameters defined in the config file as shown: '+ str(e)
+ ' or there is no need for model training.')
self.has_training = False
try:
self.raw_data_path = os.path.join(self.data_path, config['end2end_training']['input_data_path'])
self.raw_data_format = config['end2end_training']['input_data_format']
dp_config_file = os.path.join(self.config_path, config['end2end_training']['dp_config_file'])
self.dp_framework = config['end2end_training']['framework']
self.train_framework = config['end2end_training']['framework']
self.read_data_processing_steps(dp_config_file)
self.identify_dp_engine()
self.train_data_path=None
self.train_data_format=None
train_config_file = os.path.join(self.config_path, config['end2end_training']['train_config_file'])
self.read_training_configs(train_config_file)
self.test_backend = config['end2end_training']['test_backend']
try:
self.ray_params = config['end2end_training']['ray_params']
except:
self.ray_params = None
self.in_memory = True
except Exception as e:
print("Failed to read end2end training configurations. This is either due to wrong parameters defined in the config file as shown: "+ str(e)
+ " or there is no need for End-to-End training.")
self.in_memory = False
self.cluster_engine = None
def identify_dp_engine(self):
if self.dp_framework == 'pandas' and self.is_multi_nodes == False:
self.dp_engine = 'pandas'
elif self.dp_framework == 'pandas' and self.is_multi_nodes == True:
self.dp_engine = 'modin'
else:
self.df_engine = 'spark'
def prepare_env(self):
if self.is_multi_nodes:
print("enter distributed mode...")
if self.has_dp and self.has_training:
if self.dp_framework == 'spark' and self.train_framework == 'spark':
self.cluster_engine='spark'
else:
self.cluster_engine='ray'
elif self.has_dp and not self.has_training:
if self.dp_framework == 'spark':
self.cluster_engine='spark'
else:
self.cluster_engine='ray'
elif not self.has_dp and self.has_training:
if self.train_framework == 'spark':
self.cluster_engine='spark'
else:
self.cluster_engine='ray'
elif self.in_memory:
if self.dp_framework == 'spark':
self.cluster_engine='spark'
else:
self.cluster_engine='ray'
else:
print("Check your workflow config file. Program End.")
sys.exit()
if self.cluster_engine == 'ray':
print("initializing ray cluster...")
import ray
ray.init('auto', runtime_env={'env_vars': {'__MODIN_AUTOIMPORT_PANDAS__': '1'}}, log_to_driver=False)
if self.cluster_engine == 'spark':
raise NotImplementedError("spark cluster engine is currently not supported in config-style usage of the workflow.")
else:
print("enter single-node mode...")
if not self.has_dp and not self.has_training and not self.in_memory:
print("Program End.")
sys.exit()
def read_data_processing_steps(self, dp_config_file):
with open(dp_config_file, 'r') as file:
dp_steps = yaml.safe_load(file)
self.pre_splitting_steps = dp_steps['pre_splitting_transformation']
self.data_splitting_rule = dp_steps['data_splitting']
self.post_splitting_steps = dp_steps['post_splitting_transformation']
def read_training_configs(self, train_config_file):
with open(train_config_file, 'r') as file:
train_configs = yaml.safe_load(file)
self.train_data_spec = train_configs['data_spec']
try:
self.hpo_spec = train_configs['hpo_spec']
self.hpo_needed = True
except:
self.hpo_spec = None
self.hpo_needed = False
print("no need for HPO")
try:
self.train_model_spec = train_configs['model_spec']
self.hpo_needed = False
except:
self.train_model_spec = None
self.hpo_needed = True
print("no need for training")
if self.hpo_spec is None and self.train_model_spec is None:
print("none of the hpo_spec and model_spec is specified. Program End.")
sys.exit()
elif self.hpo_spec is not None and self.train_model_spec is not None:
print("Pls specify either hpo_spec or model_spec. Both are not accepted. Program End.")
sys.exit()
def read_raw_data(self):
print('reading raw data...')
if self.raw_data_format == 'csv':
self.data = read_csv_files(self.raw_data_path, engine=self.dp_engine)
def read_train_data(self):
print('reading training data...')
if self.train_data_format == 'csv':
self.data = read_csv_files(self.train_data_path, engine='pandas')
def pre_splitting_transform(self):
print("transform pre-splitting data...")
if self.dp_engine != 'spark':
from src.preprocessing.pandas.pre_splitting_transformation import PreSplittingTransformer
pre_splitting_transformer = PreSplittingTransformer(self.data, self.pre_splitting_steps, self.dp_engine)
self.data = pre_splitting_transformer.process()
else:
raise NotImplementedError("currently only pandas-based data preprocessing is supported")
def split_data(self):
print('splitting data...')
if self.dp_engine != 'spark':
if self.dp_engine == 'modin':
import modin.pandas as pd
elif self.dp_engine == 'pandas':
import pandas as pd
from src.preprocessing.pandas.data_splitting import DataSplitter
data_splitter = DataSplitter(self.data, self.data_splitting_rule)
self.train_data, self.test_data = data_splitter.process()
self.data = None
else:
raise NotImplementedError("currently only pandas-based data preprocessing is supported")
def post_splitting_transform(self):
print("transform pre-splitting data...")
if self.dp_engine != 'spark':
from src.preprocessing.pandas.post_splitting_transformation import PostSplittingTransformer
pre_splitting_transformer = PostSplittingTransformer(self.train_data, self.test_data, self.post_splitting_steps, self.dp_engine)
self.train_data, self.test_data = pre_splitting_transformer.process()
else:
raise NotImplementedError("currently only pandas-based data preprocessing is supported")
def save_processed_data(self):
print('saving data...')
if self.dp_engine != 'spark':
if self.dp_engine == 'modin':
import modin.pandas as pd
else:
import pandas as pd
if self.processed_data_format == 'csv':
data = pd.concat([self.train_data, self.test_data])
data.to_csv(self.processed_data_path+'/processed_data.csv', index=False)
print(f'data saved under the path {self.processed_data_path}/processed_data.csv')
else:
raise NotImplementedError("currently only pandas-based data preprocessing is supported")
def train_model(self, df):
print('start training models soon...')
if self.train_framework != 'spark':
from src.training.pandas.model_training import Trainer
trainer = Trainer(self.train_data_spec, df, self.train_model_spec, self.test_backend, self.in_memory, self.tmp_path, self.worker_ips, self.ray_params, self.hpo_spec)
if self.hpo_needed:
trainer.run_hpo()
else:
trainer.process()
trainer.save_model()
else:
raise NotImplementedError('currently only pandas-based model training is supported')
def clean_up(self):
if self.cluster_engine is None:
pass
elif self.cluster_engine == 'ray':
import ray
ray.shutdown()
else:
raise NotImplementedError('spark engine is to be implemented!')
def get_model_dir():
training_task = os.getenv('training_task', 'default_task')
training_mode = os.getenv('training_mode', 'xgb')
if training_mode == 'xgb':
suffix = 'xgb'
elif training_mode == 'xgb_gnn':
suffix = 'final_xgb'
else:
suffix = 'xgboost'
model_save_path = f"/MODELS/{training_task}_{suffix}/1"
os.makedirs(model_save_path, exist_ok=True)
return model_save_path
os.environ['MODEL_DIR'] = get_model_dir()
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--config-file",
required=True,
type=str,
help="speficy the config file name")
parser.add_argument(
"--mode",
required=True,
type=int,
help="use 1 for docker, 0 for bare-metal")
args, _ = parser.parse_known_args()
wf_processor = WFProcessor(args.config_file, args.mode)
start = time.time()
wf_processor.prepare_env()
print("prepare env took %.1f seconds" % ((time.time()-start)))
if wf_processor.has_dp:
dp_start = time.time()
start = time.time()
wf_processor.read_raw_data()
print("dp read data took %.1f seconds" % ((time.time()-start)))
start = time.time()
wf_processor.pre_splitting_transform()
print("dp transform pre-splitting data took %.1f seconds" % ((time.time()-start)))
start = time.time()
wf_processor.split_data()
print("dp split data took %.1f seconds" % ((time.time()-start)))
start = time.time()
wf_processor.post_splitting_transform()
print("dp transform post-splitting data took %.1f seconds" % ((time.time()-start)))
start = time.time()
wf_processor.save_processed_data()
print("dp save data took %.1f seconds" % ((time.time()-start)))
print("data preprocessing took %.1f seconds" % ((time.time()-dp_start)))
if wf_processor.has_training:
train_start = time.time()
wf_processor.read_train_data()
wf_processor.train_model(wf_processor.data)
print("training took %.1f seconds" % ((time.time()-train_start)))
if wf_processor.in_memory:
start = time.time()
wf_processor.read_raw_data()
wf_processor.pre_splitting_transform()
wf_processor.split_data()
wf_processor.post_splitting_transform()
if wf_processor.dp_engine == 'modin':
import modin.pandas as pd
elif wf_processor.dp_engine == 'pandas':
import pandas as pd
else:
raise ImportError("currently only pandas-based data preprocessing is supported")
df = pd.concat([wf_processor.train_data, wf_processor.test_data])
wf_processor.train_model(df)
print("end2end training took %.1f seconds" % ((time.time()-start)))
wf_processor.clean_up()
print('The whole workflow processing took %.1f seconds'%(time.time()-very_start))