-
Notifications
You must be signed in to change notification settings - Fork 70
/
preproccess_binclass.py
204 lines (176 loc) · 7.43 KB
/
preproccess_binclass.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
import pandas as pd
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
import glob
import argparse
import os
from shutil import copyfile, move
from pathlib import Path
# https://github.com/matplotlib/mpl_finance
from mpl_finance import candlestick2_ochl, volume_overlay
def isnan(value):
try:
import math
return math.isnan(float(value))
except:
return False
def removeOutput(finput):
if(Path(finput)).is_file():
os.remove(finput)
def main():
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('-i', '--input',
help='a csv file of stock data', required=True)
parser.add_argument('-l', '--seq_len',
help='num of sequence length', default=20)
parser.add_argument('-lf', '--label_file',
help='a label_file')
parser.add_argument('-d', '--dimension',
help='a dimension value', type=int, default=48)
parser.add_argument('-t', '--dataset_type',
help='training or testing datasets')
parser.add_argument('-m', '--mode',
help='mode of preprocessing data', required=True)
parser.add_argument('-v', '--use_volume',
help='combine with volume.', default=False)
args = parser.parse_args()
if args.mode == 'ohlc2cs':
ohlc2cs(args.input, args.seq_len, args.dataset_type,
args.dimension, args.use_volume)
if args.mode == 'createLabel':
createLabel(args.input, args.seq_len)
if args.mode == 'img2dt':
image2dataset(args.input, args.label_file)
if args.mode == 'countImg':
countImage(args.input)
def image2dataset(input, label_file):
# python preprocess.py -m img2dt -i dataset/5/img -lf FTSE_label_5.txt
label_dict = {}
with open(label_file) as f:
for line in f:
(key, val) = line.split(',')
# print("adding {} with key {}".format(val.rstrip(), key))
label_dict[key] = val.rstrip()
# print(label_dict)
# print(list(label_dict.values())[list(label_dict.keys()).index('FTSE-80')])
path = "{}/{}".format(os.getcwd(), input)
print(path)
for filename in os.listdir(path):
# print(filename)
# print(os.getcwd())
if filename is not '':
for k, v in label_dict.items():
splitname = filename.split("_")
f, e = os.path.splitext(filename)
# print("[DEBUG] - {}".format(splitname))
newname = "{}_{}".format(splitname[0], splitname[1])
if newname == k:
# print("{} same with {} with v {}".format(filename, k, v))
new_name = "{}{}.png".format(v, f)
os.rename("{}/{}".format(path, filename),
"{}/{}".format(path, new_name))
break
folders = ['1', '0']
for folder in folders:
if not os.path.exists("{}/classes/{}".format(path, folder)):
os.makedirs("{}/classes/{}".format(path, folder))
for filename in os.listdir(path):
if filename is not '':
# print(filename[:1])
if filename[:1] == "1":
move("{}/{}".format(path, filename),
"{}/classes/1/{}".format(path, filename))
elif filename[:1] == "0":
move("{}/{}".format(path, filename),
"{}/classes/0/{}".format(path, filename))
def createLabel(fname, seq_len):
# python preprocess.py -m createLabel -l 20 -i stockdatas/EWT_training5.csv
print("Creating label . . .")
# remove existing label file
filename = fname.split('/')
# print("{} - {}".format(filename[0], filename[1][:-4]))
removeOutput("{}_label_{}.txt".format(filename[1][:-4], seq_len))
df = pd.read_csv(fname, parse_dates=True, index_col=0)
df.fillna(0)
df.reset_index(inplace=True)
df['Date'] = df['Date'].map(mdates.date2num)
for i in range(0, len(df)):
c = df.ix[i:i + int(seq_len), :]
starting = 0
endvalue = 0
label = ""
if len(c) == int(seq_len)+1:
starting = c["Close"].iloc[-2]
endvalue = c["Close"].iloc[-1]
# print(f'endvalue {endvalue} - starting {starting}')
if endvalue > starting:
label = 1
else:
label = 0
with open("{}_label_{}.txt".format(filename[1][:-4], seq_len), 'a') as the_file:
the_file.write("{}-{},{}".format(filename[1][:-4], i, label))
the_file.write("\n")
print("Create label finished.")
def countImage(input):
num_file = sum([len(files) for r, d, files in os.walk(input)])
num_dir = sum([len(d) for r, d, files in os.walk(input)])
print("num of files : {}\nnum of dir : {}".format(num_file, num_dir))
def ohlc2cs(fname, seq_len, dataset_type, dimension, use_volume):
# python preprocess.py -m ohlc2cs -l 20 -i stockdatas/EWT_testing.csv -t testing
print("Converting olhc to candlestick")
symbol = fname.split('_')[0]
symbol = symbol.split('/')[1]
print(symbol)
path = "{}".format(os.getcwd())
# print(path)
if not os.path.exists("{}/dataset/{}_{}/{}/{}".format(path, seq_len, dimension, symbol, dataset_type)):
os.makedirs("{}/dataset/{}_{}/{}/{}".format(path,
seq_len, dimension, symbol, dataset_type))
df = pd.read_csv(fname, parse_dates=True, index_col=0)
df.fillna(0)
plt.style.use('dark_background')
df.reset_index(inplace=True)
df['Date'] = df['Date'].map(mdates.date2num)
for i in range(0, len(df)):
# ohlc+volume
c = df.ix[i:i + int(seq_len) - 1, :]
if len(c) == int(seq_len):
my_dpi = 96
fig = plt.figure(figsize=(dimension / my_dpi,
dimension / my_dpi), dpi=my_dpi)
ax1 = fig.add_subplot(1, 1, 1)
candlestick2_ochl(ax1, c['Open'], c['Close'], c['High'],
c['Low'], width=1,
colorup='#77d879', colordown='#db3f3f')
ax1.grid(False)
ax1.set_xticklabels([])
ax1.set_yticklabels([])
ax1.xaxis.set_visible(False)
ax1.yaxis.set_visible(False)
ax1.axis('off')
# create the second axis for the volume bar-plot
# Add a seconds axis for the volume overlay
if use_volume:
ax2 = ax1.twinx()
# Plot the volume overlay
bc = volume_overlay(ax2, c['Open'], c['Close'], c['Volume'],
colorup='#77d879', colordown='#db3f3f', alpha=0.5, width=1)
ax2.add_collection(bc)
ax2.grid(False)
ax2.set_xticklabels([])
ax2.set_yticklabels([])
ax2.xaxis.set_visible(False)
ax2.yaxis.set_visible(False)
ax2.axis('off')
pngfile = 'dataset/{}_{}/{}/{}/{}-{}.png'.format(
seq_len, dimension, symbol, dataset_type, fname[11:-4], i)
fig.savefig(pngfile, pad_inches=0, transparent=False)
plt.close(fig)
# normal length - end
print("Converting olhc to candlestik finished.")
if __name__ == '__main__':
main()