def calc_target(start_date,
end_date,
type,
ticker1,
lagLst=lagLst,
periodLst=periodLst,
filedir=filedir,
database='db_corr'):
corr = corrlab.corrAna(filedir=filedir,
start_date=start_date,
end_date=end_date,
type=type)
conn = MySQLdb.connect(host='localhost', user='******', passwd='hhui123456')
cursor = conn.cursor()
conn.select_db(database)
dayLst = corr.generateDayLst()
for day in dayLst: # 时间跨度为1天
data = corr.concatdata([day])
symbol1 = corr.symbolDict[day][ticker1[:2]]
for lag in lagLst:
for period in periodLst:
res = pd.DataFrame()
temp = data.copy()
shifted = temp[ticker1].shift(-int(lag[:-1]), 's')
align_base = corr.get_align_base(data)
_, align_shifted = align_base.align(shifted,
join='left',
axis=0)
temp[ticker1] = align_shifted.values
temp = corr.sampledata(temp, period=period)
temp.fillna(method='ffill', inplace=True)
temp.fillna(method='bfill', inplace=True)
temp_corr = temp.corr().sort_index()
res = pd.concat([res, temp_corr[ticker1]])
res.rename(columns={0: day}, inplace=True)
res.fillna(-2, inplace=True)
for ticker2 in temp_corr.index.values:
corr_value = res[day][ticker2]
ticker2 = ticker2.split('_')[0]
symbol2 = corr.symbolDict[day][ticker2[:2]]
cursor.execute(
"""REPLACE INTO tb_corr(
start_date,
end_date,
ticker1,
symbol1,
ticker2,
symbol2,
type,
period,
lag,
corr)
VALUES (
'%s', '%s','%s','%s','%s','%s','%d','%d','%d','%.6f'
)
""" %
(day, day, ticker1, symbol1, ticker2, symbol2, type,
int(period[:-1]), int(lag[:-1]), corr_value))
conn.commit()
def __init__(self,train_start, train_end, test_start, test_end, period, lag, target, type = 0, filedir ='/hdd/ctp/day/'):
self.train_start = train_start
self.train_end = train_end
self.test_start = test_start
self.test_end = test_end
self.period = period
self.lag = lag
self.target = target
self.type = type
self.filedir = filedir
self.corr = corrlab.corrAna(filedir=self.filedir, start_date=self.train_start, end_date=self.train_end, type=self.type)
def calc_target_week(start_date,
end_date,
type,
ticker1,
lagLst=lagLst,
periodLst=periodLst,
filedir=filedir,
database='db_corr'):
'''day duration is 5 days'''
corr = corrlab.corrAna(filedir=filedir,
start_date=start_date,
end_date=end_date,
type=type)
conn = MySQLdb.connect(host='localhost', user='******', passwd='hhui123456')
cursor = conn.cursor()
conn.select_db(database)
dayLst = corr.generateDayLst()
length = len(dayLst)
for i in range(5, length):
lst = dayLst[i - 5:i]
print 'processing ', lst[0] + '-' + lst[-1]
data = corr.concatdata(lst)
symbol1 = corr.symbolDict[lst[0] + '-' + lst[-1]][ticker1[:2]]
for lag in lagLst:
for period in periodLst:
res = pd.DataFrame()
temp = data.copy()
shifted = temp[ticker1].shift(-int(lag[:-1]), 's')
align_base = corr.get_align_base(data)
_, align_shifted = align_base.align(shifted,
join='left',
axis=0)
temp[ticker1] = align_shifted.values
temp = corr.sampledata(temp, period=period)
temp.fillna(method='ffill', inplace=True)
temp.fillna(method='bfill', inplace=True)
temp_corr = temp.corr().sort_index()
res = pd.concat([res, temp_corr[ticker1]])
res.rename(columns={0: lst[0] + '-' + lst[-1]}, inplace=True)
res.fillna(-2, inplace=True)
for ticker2 in temp_corr.index.values:
corr_value = res[(lst[0] + '-' + lst[-1])][ticker2]
ticker2 = ticker2.split('_')[0]
symbol2 = corr.symbolDict[lst[0] + '-' +
lst[-1]][ticker2[:2]]
cursor.execute(
"""INSERT INTO tb_corr(
start_date,
end_date,
ticker1,
symbol1,
ticker2,
symbol2,
type,
period,
lag,
corr)
VALUES (
'%s', '%s','%s','%s','%s','%s','%d','%d','%d','%.6f'
)
""" %
(lst[0], lst[-1], ticker1, symbol1, ticker2, symbol2,
type, int(period[:-1]), int(lag[:-1]), corr_value))
conn.commit()
start_date = '20171101'
end_date = '20171215'
type = 0 # 1 for aggravated, 0 for rolling, 2 for both
ticker1 = 'ru0'
lagLst = ['1s', '5s', '10s', '30s', '60s']
periodLst = ['1s', '5s', '10s', '30s', '60s']
outputdir = u'/home/hui/Documents/corr output/'
typelst = [
'noble', 'nonferrous', 'black', 'farm', 'chemical', 'futures', 'loan'
]
analst = ['ru', 'zn', 'rb', 'jm', 'j1'] #appointed analyst
corr = corrlab.corrAna(filedir=filedir,
start_date=start_date,
end_date=end_date,
type=type)
dayLst = corr.generateDayLst()
entire = pd.DataFrame()
entire_1s, entire_5s, entire_10s = pd.DataFrame(), pd.DataFrame(
), pd.DataFrame()
# for day in dayLst:
# lst = []
# lst.append(day)
#
# data = corr.concatdata(lst)
# print (day, 'calculate done')
# target = corr.getsymbol(data, ticker1)
# one_sec = corr.sampledata(data,period = '1s')