-
Notifications
You must be signed in to change notification settings - Fork 0
/
utils.py
179 lines (135 loc) · 6.82 KB
/
utils.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
from typing import Generator, List, Tuple
import pandas as pd
import numpy as np
import os
import shutil
import torch
from constants import DATA_COLUMNS, STATES, SAMPLING_RATE
def list_experiment_files(directory: str, prefix: str, extension:str='.csv') -> List[str]:
files = []
for experiment in os.listdir(directory):
path = os.path.join(directory, experiment)
for file in os.listdir(path):
if file.startswith(prefix) and file.endswith(extension):
files.append(os.path.join(path, file))
return files
def list_files(directory: str, prefix: str, extension:str='.csv') -> List[str]:
return [os.path.join(directory, file) for file in os.listdir(directory) if file.startswith(prefix) and file.endswith(extension)]
def load_data(filename) -> pd.DataFrame:
return pd.read_csv(filename, sep=',')
class Experiment:
def __init__(self, path: str) -> None:
self. filename = path.split('/')[-1]
self.data = load_data(path).iloc[:,16:56]
#self.annotations
self.trash = pd.DataFrame()
self.boredom = pd.DataFrame()
self.flow = pd.DataFrame()
self.frustration = pd.DataFrame()
self.survey = pd.DataFrame()
self.neutral = pd.DataFrame()
def __getitem__(self, key: str) -> pd.DataFrame:
return getattr(self, key)
def __setitem__(self, key: str, value: pd.DataFrame) -> None:
setattr(self, key, value)
def __iter__(self):
return iter(['trash', *STATES])
def getStates(experiment: Experiment, experiment_cycles: List[Tuple[int, str]]):
state_begin = 0
for state_end, state_name in experiment_cycles:
state_end += state_begin
# regard margin and annotations
experiment[state_name] = pd.concat([experiment[state_name], experiment.data.iloc[state_begin:state_end]], axis=0)
state_begin = state_end
def shuffleData(data: pd.DataFrame, chunk_size: int):
chunks = [data[i: i+ chunk_size] for i in range(0, len(data), chunk_size)]
np.random.shuffle(chunks)
return pd.concat(chunks, axis=0)
def splitDataAndSave(experiment: Experiment, training_size: float, testing_size: float, validation_size: float, training_path: str, testing_path: str, validation_path: str):
for state in experiment:
if state == 'trash':
continue
chunk_number = SAMPLING_RATE * 4 # 4 seconds
chunk_size = len(experiment[state]) // chunk_number
chunks = [experiment[state][i: i+ chunk_size] for i in range(0, len(experiment[state]), chunk_size)]
np.random.shuffle(chunks)
training = pd.concat(chunks[:int(training_size * chunk_number)], axis=0)
testing = pd.concat(chunks[int(training_size * chunk_number):int((training_size + testing_size) * chunk_number)], axis=0)
training.to_csv(os.path.join(training_path, state + '_' + experiment.filename), index=False)
testing.to_csv(os.path.join(testing_path, state + '_' + experiment.filename), index=False)
try:
validation = pd.concat(chunks[int((training_size + testing_size) * chunk_number):], axis=0)
validation.to_csv(os.path.join(validation_path, state + '_' + experiment.filename), index=False)
except:
print('Validation set won\'t be created...' )
# find all paths for training and testing
# import 4 dataframes from each state
# add 'boring', 'flow', 'frustration', 'neutral' columns and fill them with 0s and 1
# merge all dataframes into one, split into 1-second chunks and shuffle it
# yield the output
def list_files_for_states(directory: str, states: List[str]) -> dict[str, List[str]]:
files = {}
for state in states:
files[state] = list_files(directory, state)
return files
def load_data_for_learning(filenames: dict[str, List[str]], iteration) -> pd.DataFrame:
data = {}
states = filenames.keys()
for state in states:
data[state] = load_data(filenames[state][iteration])
data[state][state] = 1
training_data = pd.concat([data[state] for state in states], axis=0)
training_data.fillna(int(0), inplace=True)
# training_data = shuffleData(training_data, SAMPLING_RATE)
return training_data
def getTrainingRowNum(states: List[str], training_dir: str='training') -> int:
# it should also take bach size into account, because we drop $(datasets % batch_size) rows, but for now "-5" will do
training_filenames = list_files_for_states(training_dir, states)
datasets_num = len(training_filenames[states[0]])
sum = 0
for i in range(datasets_num):
sum += len(load_data_for_learning(training_filenames, i))
return sum
def prepareXY(data: pd.DataFrame) -> Tuple[torch.Tensor, torch.Tensor]:
# x.shape = (batch_num, SAMPLE_RATE, 40)
# y.shape = (batch_num, 4)
data = data[: (len(data) // SAMPLING_RATE) * SAMPLING_RATE]
tensor = torch.tensor(data.values, dtype=torch.float32)
training = tensor.view(-1, SAMPLING_RATE, tensor.shape[1])
x = training[:, :, :len(DATA_COLUMNS)]
y = training[:, 0, len(DATA_COLUMNS):]
return x, y
def lazy_load_training_data(states: List[str], training_dir: str='training'):
training_filenames = list_files_for_states(training_dir, states)
datasets_num = len(training_filenames[states[0]]) # assuming that all states have the same number of datasets
for i in range(datasets_num):
training_data = load_data_for_learning(training_filenames, i)
training_data = shuffleData(training_data, SAMPLING_RATE)
yield prepareXY(training_data)
def load_training_data(states: List[str], training_dir: str='training'):
training_filenames = list_files_for_states(training_dir, states)
datasets_num = len(training_filenames[states[0]])
data_list = []
for i in range(datasets_num):
data_list.append(load_data_for_learning(training_filenames, i))
training_data = pd.concat(data_list, axis=0)
training_data = shuffleData(training_data, SAMPLING_RATE)
return prepareXY(training_data)
def load_testing_data(states: List[str], testing_dir: str='testing'):
# read all files at once
testing_filenames = list_files_for_states(testing_dir, states)
datasets_num = len(testing_filenames[states[0]]) # assuming that all states have the same number of datasets
data_list = []
for i in range(datasets_num):
data_list.append(load_data_for_learning(testing_filenames, i))
training_data = pd.concat(data_list, axis=0)
training_data = shuffleData(training_data, SAMPLING_RATE)
# x = training_data.iloc[:, :len(DATA_COLUMNS)]
# y = training_data.iloc[:, len(DATA_COLUMNS):]
x, y = prepareXY(training_data)
return x, y
def clear_directories(dirs: List[str]):
for dir in dirs:
if os.path.exists(dir):
shutil.rmtree(dir)
os.makedirs(dir)