def __getSimilarStock__(self):
result_dis = []
result_dt = []
result_ts = []
stock_list = sd.get_similar_industry_stock(self.ts_code)
x_data = sd.get_daily_bar(self.ts_code, self.__getx_data_start_time_point__(), self.prediction_time_point)
if x_data.empty:
self.__error__ = True
return None
else:
x = np.array(x_data['pct_change'])
ob_length = len(x)
for index, row in stock_list.iterrows():
stock = row['ts_code']
if stock != self.ts_code:
y_data = sd.get_daily_bar(stock, self.__gety_data_start_time_point__(),
self.__gety_data_end_time_point__())
if not y_data.empty:
time_point_list = np.array(y_data['trade_date'])
y = np.array(y_data['pct_change'])
for i in range(0, int(len(y) / ob_length)):
y_t = y[i * ob_length:(i + 1) * ob_length]
distance, path = similarity_calculation(x, y_t, dist=euclidean)
result_dis.append(distance)
result_dt.append(time_point_list[(i + 1) * ob_length - 1])
result_ts.append(stock)
result = pd.DataFrame({'ts_code': result_ts, 'distance': result_dis, 'time_point': result_dt})
return result.sort_values(by='distance', axis=0, ascending=True)
def __getSimilarStock__(self):
result_dis = []
result_sdt = []
result_edt = []
x_data = sd.get_daily_bar(self.ts_code, self.__getx_data_start_time_point__(), self.prediction_time_point)
if x_data.empty:
self.__error__ = True
return None
else:
x = np.array(x_data['pct_change'])
ob_length = len(x)
y_data = sd.get_daily_bar(self.ts_code, self.__gety_data_start_time_point__(),
self.__gety_data_end_time_point__())
if not y_data.empty:
time_point_list = np.array(y_data['trade_date'])
y = np.array(y_data['pct_change'])
for i in range(0, len(y) - ob_length + 1):
y_t = y[i:i + ob_length]
distance, path = similarity_calculation(x, y_t, dist=euclidean)
result_dis.append(distance)
result_sdt.append(time_point_list[i])
result_edt.append(time_point_list[i + ob_length - 1])
result = pd.DataFrame(
{'distance': result_dis, 'start_time_point': result_sdt, 'end_time_point': result_edt})
return result.sort_values(by='distance', axis=0, ascending=True)
def __count_quota__(self):
result = {}
for index, row in self.__similar_stock__.iterrows():
stock = row['ts_code']
data = sd.get_daily_bar(stock, self.__get_data_start_time_point__(), self.prediction_time_point)
if not data.empty:
data['pct_mark'] = data.apply(self.__mark__, axis=1)
x = np.array(data['pct_mark'])
result.setdefault(stock, float(len(x[(x != 0)])) / float(len(x)))
self.__activity__ = result
def __count_quota__(self):
trend_line_result = {}
for index, row in self.__similar_stock__.iterrows():
stock = row['ts_code']
time = row['time_point']
e_t = datetime.strptime(time, "%Y%m%d") + timedelta(days=self.__ob_window__)
data = sd.get_daily_bar(stock, time, datetime.strftime(e_t, "%Y%m%d"))
if not data.empty:
x = list(data['close'])
trend_line_result.setdefault(stock, x)
self.__trend_line__ = trend_line_result
def __getSimilarStock__(self):
result_dis = []
result_ts = []
stock_list = sd.get_similar_industry_stock(self.ts_code)
x_data = sd.get_daily_bar(self.ts_code, self.__get_data_start_time_point__(), self.prediction_time_point)
if x_data.empty:
self.__error__ = True
return None
else:
x = np.array(x_data['pct_change'])
for index, row in stock_list.iterrows():
stock = row['ts_code']
if stock != self.ts_code:
y_data = sd.get_daily_bar(stock, self.__get_data_start_time_point__(),
self.prediction_time_point)
if not y_data.empty:
y = np.array(y_data['pct_change'])
distance, path = similarity_calculation(x, y, dist=euclidean)
result_dis.append(distance)
result_ts.append(stock)
result = pd.DataFrame({'ts_code': result_ts, 'distance': result_dis})
return result.sort_values(by='distance', axis=0, ascending=True)
def __count_quota__(self):
trend_line_result = {}
for index, row in self.__similar_stock__.iterrows():
time = row['end_time_point']
s_t = row['start_time_point']
e_t = datetime.strptime(time, "%Y%m%d") + timedelta(days=self.__ob_window__)
data = sd.get_daily_bar(self.ts_code, s_t, datetime.strftime(e_t, "%Y%m%d"))
if not data.empty:
x = list(data['close'])
t = list(data['trade_date'])
trend_line_result.setdefault("close", x)
trend_line_result.setdefault("timeline", t)
self.__trend_line__ = trend_line_result
def __count_quota__(self):
first_pct_change = []
second_pct_change = []
third_pct_change = []
for index, row in self.__similar_stock__.iterrows():
stock = row['ts_code']
time = row['time_point']
e_t = datetime.strptime(time, "%Y%m%d") + timedelta(days=30)
data = sd.get_daily_bar(stock, time, datetime.strftime(e_t, "%Y%m%d"))
if (not data.empty) and len(data)>=3:
x = np.array(data['pct_change'])
first_pct_change.append(x[0])
second_pct_change.append(x[1])
third_pct_change.append(x[2])
self.__first_pct_change__ = np.array(first_pct_change)
self.__second_pct_change__ = np.array(second_pct_change)
self.__third_pct_change__ = np.array(third_pct_change)