def _load_industry_weight(self):
ret = pd.DataFrame()
dates = WIND_DATA_PROVIDER.biz_days_list(start_date=self.start_date,
end_date=self.end_date,
freq=self.freq)
extra_params = self._check_industry_params(self.factor_name)
for date in dates:
date = date_convert_2_str(date)
index_info = WIND_DATA_PROVIDER.get_universe(self.sec_id,
date=date,
output_weight=True)
class_info = WIND_DATA_PROVIDER.query_data(
api='w.wsd',
# sec_id=index_info[1],
sec_id=index_info.index.tolist(),
indicator='indexcode_sw',
extra_params=extra_params,
start_date=date,
end_date=date)
industry_weight = pd.DataFrame(data={
'sec_id': index_info.index,
'class_id': class_info.Data[0],
'sec_weight': index_info['weight']
},
index=index_info.index)
tmp = industry_weight.groupby('class_id').sum().T
tmp.index = [date]
tmp = WIND_QUERY_HELPER.convert_2_multi_index(tmp) \
if self.output_data_format == OutputFormat.MULTI_INDEX_DF else tmp
ret = ret.append(tmp)
return ret
def _retrieve_data(self, main_params, extra_params, output_data_format):
output_data = pd.DataFrame()
api = main_params[Header.API]
if api == 'w.wsi':
merged_extra_params = self._merge_query_params(extra_params, date=self.end_date)
raw_data = WIND_DATA_PROVIDER.query_data(api=api,
sec_id=self.sec_id,
indicator=main_params[Header.INDICATOR],
extra_params=merged_extra_params,
start_date=self.start_date,
end_date=self.end_date)
multi_factors = True if extra_params[Header.MULTIFACTORS] == 'Y' else False
output_data = WIND_QUERY_HELPER.reformat_wind_data(raw_data=raw_data, date=self.end_date,
multi_factors=multi_factors)
else:
dates = WIND_DATA_PROVIDER.biz_days_list(start_date=self.start_date,
end_date=self.end_date,
freq=self.freq)
for fetch_date in dates:
if not pd.isnull(extra_params[Header.REPORTADJ]):
date = WIND_QUERY_HELPER.latest_report_date(fetch_date)
else:
date = fetch_date
date = date_convert_2_str(date)
sec_id = WIND_DATA_PROVIDER.get_universe(self.sec_id, date=date) \
if self.is_index else self.sec_id
if api == 'w.wsd':
merged_extra_params = self._merge_query_params(extra_params, date=date)
raw_data = WIND_DATA_PROVIDER.query_data(api=api,
sec_id=sec_id,
indicator=main_params[Header.INDICATOR],
extra_params=merged_extra_params,
start_date=date,
end_date=date)
elif api == 'w.wss':
py_assert(not pd.isnull(extra_params[Header.TENOR]), ValueError,
'tenor must be given for query factor {0}'.format(self.factor_name))
merged_extra_params = self._merge_query_params(extra_params, date=date)
raw_data = WIND_DATA_PROVIDER.query_data(api=api,
sec_id=sec_id,
indicator=main_params[Header.INDICATOR],
extra_params=merged_extra_params)
else:
raise ValueError('FactorLoader._retrieve_data: unacceptable value of parameter api')
multi_factors = True if extra_params[Header.MULTIFACTORS] == 'Y' else False
tmp = WIND_QUERY_HELPER.reformat_wind_data(raw_data=raw_data,
date=fetch_date,
output_data_format=output_data_format,
multi_factors=multi_factors)
output_data = pd.concat([output_data, tmp], axis=0)
return output_data
def _retrieve_data(self, main_params, extra_params, output_data_format):
output_data = pd.DataFrame()
api = main_params[Header.API]
if api == 'w.wsq':
loop_times = int(len(self.sec_id) / self.block_size) + 1
for j in range(loop_times):
code_set = self.sec_id[j * self.block_size:(j + 1) *
self.block_size]
raw_data = WIND_DATA_PROVIDER.query_data(
api=api,
sec_id=code_set,
indicator=main_params[Header.INDICATOR])
length = len(raw_data.Data[0])
output_data = pd.concat([
output_data,
pd.DataFrame(
np.concatenate(
([raw_data.Times * length], raw_data.Data))).T
],
axis=0)
output_data.columns = ['date'] + [
field[3:] for field in main_params[Header.INDICATOR].split(',')
]
elif api == 'w.wsi':
merged_extra_params = self._merge_query_params(extra_params,
date=self.end_date)
raw_data = WIND_DATA_PROVIDER.query_data(
api=api,
sec_id=self.sec_id,
indicator=main_params[Header.INDICATOR],
extra_params=merged_extra_params,
start_date=self.start_date,
end_date=self.end_date)
multi_factors = True if extra_params[
Header.MULTIFACTORS] == 'Y' else False
output_data = WIND_QUERY_HELPER.reformat_wind_data(
raw_data=raw_data,
date=self.end_date,
multi_factors=multi_factors)
else:
dates = WIND_DATA_PROVIDER.biz_days_list(
start_date=self.start_date,
end_date=self.end_date,
freq=self.freq)
for fetch_date in dates:
if not pd.isnull(extra_params[Header.REPORTADJ]):
date = WIND_QUERY_HELPER.latest_report_date(fetch_date)
else:
date = fetch_date
date = date_convert_2_str(date)
sec_id = WIND_DATA_PROVIDER.get_universe(self.sec_id, date=date) \
if self.is_index else self.sec_id
if api == 'w.wsd':
merged_extra_params = self._merge_query_params(
extra_params, date=date)
raw_data = WIND_DATA_PROVIDER.query_data(
api=api,
sec_id=sec_id,
indicator=main_params[Header.INDICATOR],
extra_params=merged_extra_params,
start_date=date,
end_date=date)
elif api == 'w.wss':
# py_assert(not pd.isnull(extra_params[Header.TENOR]), ValueError,
# 'tenor must be given for query factor {0}'.format(self.factor_name))
merged_extra_params = self._merge_query_params(
extra_params, date=date)
raw_data = WIND_DATA_PROVIDER.query_data(
api=api,
sec_id=sec_id,
indicator=main_params[Header.INDICATOR],
extra_params=merged_extra_params)
else:
raise ValueError(
'FactorLoader._retrieve_data: unacceptable value of parameter api'
)
multi_factors = True if extra_params[
Header.MULTIFACTORS] == 'Y' else False
tmp = WIND_QUERY_HELPER.reformat_wind_data(
raw_data=raw_data,
date=fetch_date,
output_data_format=output_data_format,
multi_factors=multi_factors)
output_data = pd.concat([output_data, tmp], axis=0)
return output_data