TraFix/analysis.py at master · omerktz/TraFix · GitHub

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
import os
import csv
import logging
from dateutil.parser import parse
from utils.colored_logger_with_timestamp import init_colorful_root_logger


def analyze_all_results(logic, keys, logs, outputs):
	return map(lambda k: logic(logs[k], outputs[k]), keys)


def analyze_all_results_average(logic, keys, logs, outputs, missing_is_zero):
	results = map(lambda k: logic(logs[k], outputs[k]), keys)
	inner_keys = set(reduce(lambda x, y: x + y, map(lambda r: r.keys(), results)))
	for i in range(max(inner_keys)):
		if i not in inner_keys:
			inner_keys.add(i)
	inner_keys = sorted(inner_keys)
	for result in results:
		for ik in inner_keys:
			if ik not in result.keys():
				if missing_is_zero:
					result[ik] = 0
				else:
					result[ik] = result[ik - 1]
	final_results = {}
	for ik in inner_keys:
		data = map(lambda r: r[ik], results)
		final_results[ik] = (sum(data) / float(len(data)), min(data), max(data))
	return final_results


def get_datasets_size(folder, prefix, suffix):
	iterations = set()
	for f in os.listdir(os.path.join(folder, 'datasets')):
		if f.startswith(prefix) and f.endswith(suffix):
			iterations.add(int(f[len(prefix):-len(suffix)]))
	result = {}
	for i in iterations:
		with open(os.path.join(folder, 'datasets', '{0}{1}{2}'.format(prefix, i, suffix)), 'r') as f:
			result[i] = len(filter(lambda l: len(l.strip()) > 0, f.readlines()))
	return result


def get_timings(log, marker, until=1):
	def get_time(line):
		index = line.find(' ')
		return parse(line[index - 10: index + 8])

	result = {}
	with open(log, 'r') as f:
		lines = filter(lambda y: len(y) > 0, map(lambda x: x.strip(), f.readlines()))
		for i in range(len(lines)):
			if marker in lines[i]:
				line = lines[i][:]
				line = line[line.find(' (iteration ') + 12:]
				line = line[:line.find(')')]
				result[int(line)] = (get_time(lines[i + until]) - get_time(lines[i])).total_seconds()
	return result


def num_of_iterations(keys, logs, outputs, out):
	def logic(log, output):
		with open(log, 'r') as f:
			lines = filter(lambda l: len(l.strip()) > 0, f.readlines())
			last_line = lines[-1]
			last_line = last_line[last_line.find('%) in ') + 6:]
			last_line = last_line[:last_line.find(' ')]
			return int(last_line)

	data = analyze_all_results(logic, keys, logs, outputs)
	out.writerow(['Number of Iterations'])
	out.writerow(['', 'Raw:'] + data)
	out.writerow(['', 'Min:', min(data), '', 'Max:', max(data), '', 'Average:', sum(data) / len(data)])
	out.writerow([])


def remaining_test_size(keys, logs, outputs, out):
	def logic(log, output):
		return get_datasets_size(output, 'test', '.corpus.hl')

	data = analyze_all_results_average(logic, keys, logs, outputs, missing_is_zero=False)
	out.writerow(['Remaining test size'])
	out.writerow(['', 'Average:'])
	out.writerow(['', '', 'Raw:'] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', '', 'Ratio:'] + map(lambda k: data[k][0] / float(data[0][0]), sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow(['', '', 'Raw:'] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', '', 'Ratio:'] + map(lambda k: data[k][1] / float(data[0][1]), sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow(['', '', 'Raw:'] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow(['', '', 'Ratio:'] + map(lambda k: data[k][2] / float(data[0][2]), sorted(data.keys())))
	out.writerow([])
	out.writerow(['Successful test samples'])
	out.writerow(['', 'Average:'])
	out.writerow(['', '', 'Raw:'] + map(lambda k: data[0][0] - data[k][0], sorted(data.keys())))
	out.writerow(['', '', 'Ratio:'] + map(lambda k: (data[0][0] - data[k][0]) / float(data[0][0]), sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow(['', '', 'Raw:'] + map(lambda k: data[0][2] - data[k][2], sorted(data.keys())))
	out.writerow(['', '', 'Ratio:'] + map(lambda k: (data[0][2] - data[k][2]) / float(data[0][2]), sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow(['', '', 'Raw:'] + map(lambda k: data[0][1] - data[k][1], sorted(data.keys())))
	out.writerow(['', '', 'Ratio:'] + map(lambda k: (data[0][1] - data[k][1]) / float(data[0][1]), sorted(data.keys())))
	out.writerow([])


def train_size(keys, logs, outputs, out):
	def logic(log, output):
		return get_datasets_size(output, 'train', '.corpus.hl')

	data = analyze_all_results_average(logic, keys, logs, outputs, missing_is_zero=False)
	out.writerow(['Train set size'])
	out.writerow(['', 'Average:'])
	out.writerow([''] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow([''] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow([''] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow([])


def train_time(keys, logs, outputs, out):
	def logic(log, output):
		return get_timings(log, 'Training model ')

	data = analyze_all_results_average(logic, keys, logs, outputs, missing_is_zero=True)
	out.writerow(['Training time (seconds)'])
	out.writerow(['', 'Average:'])
	out.writerow([''] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow([''] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow([''] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow([])


def translate_time(keys, logs, outputs, out):
	def logic(log, output):
		return get_timings(log, 'Translating dataset ')

	data = analyze_all_results_average(logic, keys, logs, outputs, missing_is_zero=True)
	out.writerow(['Translation time (seconds)'])
	out.writerow(['', 'Average:'])
	out.writerow([''] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow([''] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow([''] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow([])


def evaluate_time(keys, logs, outputs, out):
	def logic(log, output):
		return get_timings(log, 'Evaluating latest results ')

	data = analyze_all_results_average(logic, keys, logs, outputs, missing_is_zero=True)
	out.writerow(['Evaluation time (seconds)'])
	out.writerow(['', 'Average:'])
	out.writerow([''] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow([''] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow([''] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow([])


def iteration_time(keys, logs, outputs, out):
	def logic(log, output):
		return get_timings(log, 'Training model ', until=4)

	data = analyze_all_results_average(logic, keys, logs, outputs, missing_is_zero=True)
	out.writerow(['Iteration time (seconds)'])
	out.writerow(['', 'Average:'])
	out.writerow([''] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow([''] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow([''] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow([])


def train_epochs(keys, logs, outputs, out):
	def logic_last(log, output):
		iterations = set()
		for f in os.listdir(os.path.join(output, 'outputs')):
			if f.startswith('train'):
				iterations.add(int(f[len('train'):]))
		result = {}
		for i in iterations:
			with open(os.path.join(output, 'outputs', '{0}{1}'.format('train', i)), 'r') as f:
				lines = filter(lambda l: len(l.strip()) > 0, f.readlines())
				last_epoch = int(lines[-3][len('last epoch is '):])
				result[i] = last_epoch
		return result
	def logic_best(log, output):
		iterations = set()
		for f in os.listdir(os.path.join(output, 'outputs')):
			if f.startswith('train'):
				iterations.add(int(f[len('train'):]))
		result = {}
		for i in iterations:
			with open(os.path.join(output, 'outputs', '{0}{1}'.format('train', i)), 'r') as f:
				lines = filter(lambda l: len(l.strip()) > 0, f.readlines())
				best_epoch = int(lines[-2][len('best epoch is '):])
				result[i] = best_epoch
		return result

	data = analyze_all_results_average(logic_last, keys, logs, outputs, missing_is_zero=True)
	out.writerow(['Last training epoch'])
	out.writerow(['', 'Average:'])
	out.writerow([''] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow([''] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow([''] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow([])
	data = analyze_all_results_average(logic_best, keys, logs, outputs, missing_is_zero=True)
	out.writerow(['Best training epoch'])
	out.writerow(['', 'Average:'])
	out.writerow([''] + map(lambda k: data[k][0], sorted(data.keys())))
	out.writerow(['', 'Min:'])
	out.writerow([''] + map(lambda k: data[k][1], sorted(data.keys())))
	out.writerow(['', 'Max:'])
	out.writerow([''] + map(lambda k: data[k][2], sorted(data.keys())))
	out.writerow([])


analysis_funcs = [num_of_iterations, remaining_test_size, train_size, train_time, translate_time, evaluate_time,
				  iteration_time, train_epochs]


def analyze_results(input, output, funcs=analysis_funcs):
	logs = {}
	outputs = {}
	for f in os.listdir(input):
		path = os.path.join(input, f)
		if os.path.isdir(path) and f.startswith('output'):
			outputs[int(f[6:])] = path
		if os.path.isfile(path) and f.startswith('log'):
			logs[int(f[3:])] = path
	keys = set()
	for i in logs.keys():
		if i not in outputs.keys():
			logging.info('Ignoring log{0}, no matching output folder'.format(i))
		else:
			keys.add(i)
	for i in outputs.keys():
		if i not in logs.keys():
			logging.info('Ignoring output{0}, no matching log file'.format(i))
		else:
			keys.add(i)
	with open(output, 'w') as fout:
		csvout = csv.writer(fout)
		csvout.writerow(['Number of experiments:', len(keys)])
		csvout.writerow([])
		for f in funcs:
			f(sorted(keys), logs, outputs, csvout)


if __name__ == "__main__":
	import argparse

	parser = argparse.ArgumentParser(description="Analyze results of Active Learner")
	parser.add_argument('input', type=str, help="Folder used as input")
	parser.add_argument('output', type=str, help="Output file name")
	parser.add_argument('-v', '--verbose', action='store_const', const=True, help='Be verbose')
	parser.add_argument('--debug', action='store_const', const=True, help='Enable debug prints')

	args = parser.parse_args()
	init_colorful_root_logger(logging.getLogger(''), vars(args))

	analyze_results(input=args.input, output=args.output)