def __init__(self, filename, parent=None):
# Sets up window parent
super(SnippingWindow, self).__init__(parent)
self.main = False
# Sets main window bar title and size
self.setWindowTitle('Record Snipper')
self.resize(800, 480)
if filename == '':
self.main = True
self.all_files = ef.get_all_files('myo_data')
if not len(self.all_files):
self.closeEvent()
print('Loading %s files' % len(self.all_files))
filename = self.all_files[0]
self.all_files = self.all_files[1:]
if not filename == '':
self.load_record(filename)
print('Record Loaded')
self.all_plots()
def get_today_more_data_for_MLP(code):
'''
:param code:股票代码
:return: 今日预测的X
'''
import numpy as np
stock_data_path = "./data/stock_data/"
open_price, oneDayLine, volume, ma5, vma5, turnover, dates = load_fq_open_close_volume_ma5_vma5_turnover_from_tushare(
stock_data_path + str(code) + '_fq.csv')
if (str(code)[0] == '6'):
# 上证指数
open_index, close_index, volume_index, ma5_index, vma5_index, dates_index = load_index_open_close_volume_ma5_vma5_from_tushare(
stock_data_path + '../sh.csv')
else:
# 深证指数
open_index, close_index, volume_index, ma5_index, vma5_index, dates_index = load_index_open_close_volume_ma5_vma5_from_tushare(
stock_data_path + '../sz.csv')
daynum = 5
X_clf = []
ef = Extract_Features()
for i in range(daynum, len(dates)):
#以大盘为标准
p = dates_index.index(dates[i])
# 组装数据
X_delta = [(oneDayLine[k] - oneDayLine[k - 1]) / oneDayLine[k - 1] for k in range(i - daynum, i)] + \
[(volume[k] - volume[k - 1]) / volume[k - 1] for k in range(i - daynum, i)] + \
[turnover[k] for k in range(i - daynum, i)] + \
[(ma5[i] - ma5[i - 1]) / ma5[i - 1]] + \
[(vma5[i] - vma5[i - 1]) / vma5[i - 1]] + \
[(open_index[p] - open_index[p - 1]) / open_index[p - 1]] + \
[(close_index[p] - close_index[p - 1]) / close_index[p - 1]] + \
[(volume_index[p] - volume_index[p - 1]) / volume_index[p - 1]] + \
[(ma5_index[p] - ma5_index[p - 1]) / ma5_index[p - 1]] + \
[(vma5_index[p] - vma5_index[p - 1]) / vma5_index[p - 1]] + \
covert_bin(ef.parse_weekday(dates[i]), 3) \
# + [(ef.MoneySupply(dates[i]) - ef.MoneySupply(dates[i - 22])) / ef.MoneySupply(dates[i - 22])]
# covert_bin(ef.lunar_month(dates[i]), 5) + \
if i == 5:
X_delta[0] = X_delta[5] = 0.0 # 调整第一个百分比
X_clf.append(X_delta)
X_clf = preprocessing.MinMaxScaler().fit_transform(X_clf)
# return np.array(X_clf)
return X_clf[-1]
def get_last_N_days_data_for_MLP(code, N):
'''
'''
import numpy as np
data_path = "./data/stock_data/"
oneDayLine, date = load_data_from_tushare(data_path + str(code) + '.csv')
volumn, volumn_dates = load_volume_from_tushare(data_path + str(code) +
'.csv')
# each data consist of delta od last daynum days
daynum = 5
num = N
X_clf = []
y_clf = []
ef = Extract_Features()
for i in range(len(date) - num, len(date)):
X_delta = [oneDayLine[k] - oneDayLine[k - 1] for k in range(i - daynum, i)] + \
[volumn[k] for k in range(i - daynum, i)] + \
[float(ef.parse_weekday(date[i - 1]))] + \
[float(ef.lunar_month(date[i - 1]))] + \
[float(ef.rrr(date[i - 1]))] + \
[float(ef.MoneySupply(date[i - 1]))]
X_clf.append(X_delta)
y_clf.append([1, 0] if oneDayLine[i] -
oneDayLine[i - 1] >= 0 else [0, 1])
return np.array(X_clf), np.array(y_clf)
def get_data_for_MLP(code):
'''
'''
import numpy as np
data_path = "./data/stock_data/"
oneDayLine, date = load_data_from_tushare(data_path + str(code) + '.csv')
volumn, volumn_dates = load_volume_from_tushare(data_path + str(code) +
'.csv')
daynum = 5
X = []
y = []
ef = Extract_Features()
for i in range(daynum, len(date)):
X_delta = [oneDayLine[k] - oneDayLine[k - 1] for k in range(i - daynum, i)] + \
[volumn[k] for k in range(i - daynum, i)] + \
[float(ef.parse_weekday(date[i - 1]))] + \
[float(ef.lunar_month(date[i - 1]))] + \
[ef.rrr(date[i - 1])] + \
[ef.MoneySupply(date[i - 1])]
X.append(X_delta)
y.append([1, 0] if oneDayLine[i] - oneDayLine[i - 1] >= 0 else [0, 1])
return np.array(X), np.array(y)
def get_today_data_for_MLP(code):
'''
'''
import numpy as np
data_path = "./data/stock_data/"
oneDayLine, date = load_data_from_tushare(data_path + str(code) + '.csv')
volumn, volumn_dates = load_volume_from_tushare(data_path + str(code) +
'.csv')
daynum = 5
X = []
ef = Extract_Features()
for i in range(daynum, len(date)):
X_delta = [oneDayLine[k] - oneDayLine[k - 1] for k in range(i - daynum, i)] + \
[volumn[k] for k in range(i - daynum, i)] + \
[float(ef.parse_weekday(date[i]))] + \
[float(ef.lunar_month(date[i]))] + \
[ef.rrr(date[i])] + \
[ef.MoneySupply(date[i])]
X.append(X_delta)
X = preprocessing.MinMaxScaler().fit_transform(X)
return np.array(X[-1])
def get_last_month_more_data_for_MLP(code):
'''
'''
import numpy as np
stock_data_path = "./data/stock_data/"
open_price, oneDayLine, volume, ma5, vma5, turnover, dates = load_fq_open_close_volume_ma5_vma5_turnover_from_tushare(
stock_data_path + str(code) + '_fq.csv')
if (str(code)[0] == '6'):
# 上证指数
open_index, close_index, volume_index, ma5_index, vma5_index, dates_index = load_index_open_close_volume_ma5_vma5_from_tushare(
stock_data_path + '../sh.csv')
else:
# 深证指数
open_index, close_index, volume_index, ma5_index, vma5_index, dates_index = load_index_open_close_volume_ma5_vma5_from_tushare(
stock_data_path + '../sz.csv')
daynum = 5
num = 22
X_clf = []
ef = Extract_Features()
# for i in range(daynum, len(dates)):
# #以大盘为标准
#
# # if dates_index[p] in dates:
# # # 对齐大盘与个股的日期,得到个股的对应日期位置i
# # i = dates.index(dates_index[p])
# # else:
# # X_delta = [0 for x in range(21)]
# # X_clf.append(X_delta)
# # continue
# p = dates_index.index(dates[i])
#
# # 大单交易数据
# big_deals = get_big_deal_volume(code, dates[i])
#
# # 组装数据
# X_delta = [oneDayLine[k] - oneDayLine[k - 1] for k in range(i - daynum, i)] + \
# [volume[k] - volume[k - 1] for k in range(i - daynum, i)] + \
# [ma5[i]] + \
# [vma5[i]] + \
# [open_index[p]] + [close_index[p]] + [volume_index[p]] + [ma5_index[p]] + [
# vma5_index[p]] + \
# [big_deals] + \
# [ef.parse_weekday(dates[i])] + \
# [ef.lunar_month(dates[i])] + \
# [ef.MoneySupply(dates[i])]
# X_clf.append(X_delta)
# # y_clf.append([1, 0] if oneDayLine[i] - oneDayLine[i - 1] > 0 else [0, 1])
# X_clf = preprocessing.MinMaxScaler().fit_transform(X_clf)
# return X_clf
for i in range(daynum, len(dates)):
#以大盘为标准
p = dates_index.index(dates[i])
# 组装数据
X_delta = [(oneDayLine[k] - oneDayLine[k - 1]) / oneDayLine[k - 1] for k in range(i - daynum, i)] + \
[(volume[k] - volume[k - 1]) / volume[k - 1] for k in range(i - daynum, i)] + \
[turnover[k] for k in range(i - daynum, i)] + \
[(ma5[i] - ma5[i - 1]) / ma5[i - 1]] + \
[(vma5[i] - vma5[i - 1]) / vma5[i - 1]] + \
[(open_index[p] - open_index[p - 1]) / open_index[p - 1]] + \
[(close_index[p] - close_index[p - 1]) / close_index[p - 1]] + \
[(volume_index[p] - volume_index[p - 1]) / volume_index[p - 1]] + \
[(ma5_index[p] - ma5_index[p - 1]) / ma5_index[p - 1]] + \
[(vma5_index[p] - vma5_index[p - 1]) / vma5_index[p - 1]] + \
covert_bin(ef.parse_weekday(dates[i]), 3) \
# + [(ef.MoneySupply(dates[i]) - ef.MoneySupply(dates[i - 22])) / ef.MoneySupply(dates[i - 22])]
# covert_bin(ef.lunar_month(dates[i]), 5) + \
if i == 5:
X_delta[0] = X_delta[5] = 0.0 # 调整第一个百分比
X_clf.append(X_delta)
X_clf = preprocessing.MinMaxScaler().fit_transform(X_clf)
# return np.array(X_clf)
return X_clf