def load_data(cls, date, code_list=None):
print("Downloading data ...")
tuple_str = "("
for s in report_period_generator(period=32, date=date):
tuple_str += "'" + s + "',"
tuple_str = tuple_str[:-1]
tuple_str += ")"
sql = '''
SELECT
S_INFO_WINDCODE,
REPORT_PERIOD,
ACTUAL_ANN_DT,
STATEMENT_TYPE,
NET_PROFIT_EXCL_MIN_INT_INC AS net_income,
TOT_OPER_REV AS total_revenue,
OPER_REV AS revenue,
TOT_OPER_COST AS total_opcost,
LESS_OPER_COST AS operating_cost,
LESS_SELLING_DIST_EXP AS sale_expense,
LESS_GERL_ADMIN_EXP AS management_expense,
RD_EXPENSE AS research_expense,
LESS_FIN_EXP AS financial_expense,
OPER_PROFIT AS operating_profit
FROM
wind.ASHAREINCOME
WHERE
SUBSTR(S_INFO_WINDCODE, 1, 1) != 'A'
AND STATEMENT_TYPE in (408001000, 408004000, 408005000, 408050000)
AND REPORT_PERIOD in {0}
AND ACTUAL_ANN_DT <= {1}
'''.format(tuple_str, date)
df = db.query_by_SQL("wind", sql)
return df, date
def load_data(cls, date, code_list=None):
print("Downloading data ...")
tuple_str = "("
for s in report_period_generator(period=32, date=date):
tuple_str += "'" + s + "',"
tuple_str = tuple_str[:-1]
tuple_str += ")"
sql = '''
SELECT
S_INFO_WINDCODE,
REPORT_PERIOD,
ACTUAL_ANN_DT,
STATEMENT_TYPE,
NET_CASH_FLOWS_OPER_ACT AS operating_cashflow,
STOT_CASH_INFLOWS_OPER_ACT AS operating_cashinflow,
STOT_CASH_OUTFLOWS_OPER_ACT AS operating_cashoutflow,
STOT_CASH_INFLOWS_INV_ACT AS investment_cashinflow,
STOT_CASH_OUTFLOWS_INV_ACT AS investment_cashoutflow,
NET_CASH_FLOWS_INV_ACT AS investment_cashflow,
STOT_CASH_INFLOWS_FNC_ACT AS finance_cashinflow,
STOT_CASH_OUTFLOWS_FNC_ACT AS finance_cashoutflow,
NET_CASH_FLOWS_FNC_ACT AS finance_cashflow
FROM
wind.ASHARECASHFLOW
WHERE
SUBSTR(S_INFO_WINDCODE, 1, 1) != 'A'
AND STATEMENT_TYPE in (408001000, 408004000, 408005000, 408050000)
AND REPORT_PERIOD in {0}
AND ACTUAL_ANN_DT <= {1}
'''.format(tuple_str, date)
df = db.query_by_SQL("wind", sql)
return df, date
def load_data(cls, date, code_list=None):
print("Downloading data ...")
tuple_str = "("
for s in report_period_generator(period=32, date=date):
tuple_str += "'" + s + "',"
tuple_str = tuple_str[:-1]
tuple_str += ")"
sql = '''
SELECT
S_INFO_WINDCODE,
REPORT_PERIOD,
ACTUAL_ANN_DT,
STATEMENT_TYPE,
TOT_ASSETS AS total_assets,
TOT_SHRHLDR_EQY_EXCL_MIN_INT AS total_equities_exc_min,
TOT_SHRHLDR_EQY_INCL_MIN_INT AS total_equities_inc_min,
TOT_NON_CUR_LIAB AS noncur_liabilities,
TOT_LIAB AS total_liabilities,
LT_BORROW AS longterm_loan,
BONDS_PAYABLE AS bonds_payable,
LT_PAYABLE AS longterm_payable,
OTHER_EQUITY_TOOLS_P_SHR AS preferred_stock,
MONETARY_CAP AS cash,
TRADABLE_FIN_ASSETS AS tradable_financialasset,
NOTES_RCV AS notes_receiveable,
ACCT_RCV AS accounts_receivable,
INVENTORIES AS inventory,
FIX_ASSETS AS fixed_asset,
CONST_IN_PROG AS construction_inprogress,
INTANG_ASSETS AS intangible_asset,
R_AND_D_COSTS AS development_expenditure,
GOODWILL AS goodwill,
NOTES_PAYABLE AS notes_payable,
ACCT_PAYABLE AS accounts_payable
FROM
wind.ASHAREBALANCESHEET
WHERE
SUBSTR(S_INFO_WINDCODE, 1, 1) != 'A'
AND STATEMENT_TYPE in (408001000, 408004000, 408005000, 408050000)
AND REPORT_PERIOD in {0}
AND ACTUAL_ANN_DT <= {1}
'''.format(tuple_str, date)
df = db.query_by_SQL("wind", sql)
return df, date
def fit(cls, datas):
print("Processing data ...")
df, date = datas
df = pd.merge(df, get_listed_stocks(), on="s_info_windcode")
df.sort_values(by=[
"s_info_windcode", "report_period", "actual_ann_dt",
"statement_type"
],
inplace=True)
current_df = df.groupby(by="s_info_windcode").last()
data_series = [(dt.datetime.strftime(dt.datetime.now(), "%Y%m%d"),
list(current_df.index), current_df)]
for snapshot_date in report_period_generator(period=24, date=date):
# df_slice = df[(df["actual_ann_dt"] <= snapshot_date) & (df["report_period"] <= snapshot_date)].copy()
df_slice = df[df["report_period"] <= snapshot_date].copy()
df_slice.sort_values(by=[
"s_info_windcode", "report_period", "actual_ann_dt",
"statement_type"
],
inplace=True)
df_slice = df_slice.groupby(by="s_info_windcode").last()
data_series.append([snapshot_date, list(df_slice.index), df_slice])
data_df = pd.DataFrame(data_series,
columns=["date", "ticker_list", "report"])
for ticker in data_df.loc[0, "ticker_list"]:
level = 0
while level < 20 and ticker in data_df.loc[level + 1,
"ticker_list"]:
level += 1
data_df.loc[0, "report"].loc[ticker, "max_level"] = level
data_df.loc[0, "report"]["max_level"] = data_df.loc[
0, "report"]["max_level"].astype(int)
print("Making snapshots ...")
ttm_df = pd.DataFrame(
columns=list(data_df.loc[0, "report"].columns)[3:-1])
for ticker in data_df.loc[0, "ticker_list"]:
if data_df.loc[0, "report"].loc[ticker, "max_level"] >= 5:
shift = Quarters2LastDec(date)
ttm_df.loc[ticker, :] = data_df.loc[0, "report"].ix[ticker, 3:] \
+ data_df.loc[shift, "report"].ix[ticker, 3:] - data_df.loc[5, "report"].ix[ticker, 3:]
ttm_df = ttm_df.reset_index()
ttm_df.rename(columns={'index': 'code'}, inplace=True)
ttm_df["trade_date"] = date
ttm_df["trade_date"] = date
new_cols = list(ttm_df.columns)
new_cols.insert(0, new_cols.pop(new_cols.index("trade_date")))
ttm_df = ttm_df[new_cols]
for col in ttm_df.columns[2:]:
ttm_df[col + "_snapshots"] = None
for col in ttm_df.columns:
if col.endswith("_snapshots"):
for index in ttm_df.index:
ttm_df.set_value(index, col, dict())
ttm_df.index = ttm_df["code"]
for k in range(20, 0, -1):
snapshot_date = data_df.loc[k, "date"]
shift = Quarters2LastDec(snapshot_date)
for ticker in data_df.loc[0, "ticker_list"]:
if data_df.loc[0, "report"].loc[ticker, "max_level"] >= k + 4:
for col in data_df.loc[k + shift, "report"].columns[3:]:
ttm_df.loc[ticker, col + "_snapshots"][snapshot_date] = data_df.loc[k, "report"].loc[ticker, col] \
+ data_df.loc[k + shift, "report"].loc[ticker, col] - data_df.loc[k + 4, "report"].loc[ticker, col]
print("Converting some NaN to 0 ...")
df = ttm_df
for col in df.columns:
if col.endswith("_snapshots"):
df[col] = df[col].astype(str)
for factor in [
"net_income", "total_revenue", "revenue", "total_opcost",
"operating_cost", "sale_expense", "management_expense",
"research_expense", "financial_expense", "operating_profit"
]:
df[factor + "_snapshots"] = df[factor + "_snapshots"].apply(
lambda s: s.replace("{}", ''))
if factor not in ["net_income", "total_revenue"]:
df[factor] = df[factor].apply(lambda x: 0 if pd.isna(x) else x)
df[factor + "_snapshots"] = df[factor + "_snapshots"].apply(
lambda s: s.replace("nan", '0'))
return df
def fit(cls, datas):
df, date = datas
print("Processing data ...")
df = pd.merge(df, get_listed_stocks(), on="s_info_windcode")
df.sort_values(by=["s_info_windcode", "report_period", "actual_ann_dt", "statement_type"], inplace=True)
current_df = df.groupby(by="s_info_windcode").last()
data_series = [(dt.datetime.strftime(dt.datetime.now(), "%Y%m%d"), list(current_df.index), current_df)]
for snapshot_date in report_period_generator(period=20, date=date):
df_slice = df[(df["actual_ann_dt"] <= snapshot_date) & (df["report_period"] <= snapshot_date)].copy()
df_slice.sort_values(by=["s_info_windcode", "report_period", "actual_ann_dt", "statement_type"],
inplace=True)
df_slice = df_slice.groupby(by="s_info_windcode").last()
data_series.append([snapshot_date, list(df_slice.index), df_slice])
data_df = pd.DataFrame(data_series, columns=["date", "ticker_list", "report"])
for ticker in data_df.loc[0, "ticker_list"]:
level = 0
while level < 20 and ticker in data_df.loc[level + 1, "ticker_list"]:
level += 1
data_df.loc[0, "report"].loc[ticker, "max_level"] = level
data_df.loc[0, "report"]["max_level"] = data_df.loc[0, "report"]["max_level"].astype(int)
df = data_df.loc[0, "report"].copy()
df.drop(["report_period", "actual_ann_dt", "statement_type", "max_level"], axis=1, inplace=True)
for factor in df.columns:
df[factor + "_snapshots"] = [dict(), ] * df.shape[0]
for col in df.columns:
if col.endswith("_snapshots"):
for index in df.index:
df.set_value(index, col, dict())
print("Making snapshots ...")
for ticker in data_df.loc[0, "ticker_list"]:
max_level = data_df.loc[0, "report"].loc[ticker, "max_level"]
for factor in data_df.loc[0, "report"].columns[
3: -1]: # Exclude ["report_period", "actual_ann_dt", "statement_type", "max_level"]
for level in range(max_level, 0, -1):
df.loc[ticker, factor + "_snapshots"][data_df.loc[level, "date"]] = \
data_df.loc[level, "report"].loc[ticker, factor]
for col in df.columns:
if col.endswith("_snapshots"):
df[col] = df[col].astype(str)
print("Converting some NaN to 0 ...")
for factor in ['total_assets', 'total_equities_exc_min', 'total_equities_inc_min',
'noncur_liabilities', 'total_liabilities',
'longterm_loan', 'bonds_payable', 'longterm_payable', 'preferred_stock',
"cash", "tradable_financialasset", "notes_receiveable", "accounts_receivable",
"inventory", "fixed_asset", "construction_inprogress", "intangible_asset",
"development_expenditure", "goodwill", "notes_payable", "accounts_payable"]:
df[factor + "_snapshots"] = df[factor + "_snapshots"].apply(lambda s: s.replace("{}", ''))
if factor not in ['total_assets', 'total_equities_exc_min', 'total_equities_inc_min',
'noncur_liabilities', 'total_liabilities']:
df[factor] = df[factor].apply(lambda x: 0 if pd.isna(x) else x)
df[factor + "_snapshots"] = df[factor + "_snapshots"].apply(lambda s: s.replace("nan", '0'))
df = df.reset_index()
df.rename(columns={'s_info_windcode': 'code'}, inplace=True)
df["trade_date"] = date
new_cols = list(df.columns)
new_cols.insert(0, new_cols.pop(new_cols.index("trade_date"))) # Put trade_date into the first column
df = df[new_cols]
return df
def fit(cls, datas):
print("Processing data ...")
df, date = datas
df = pd.merge(df, get_listed_stocks(), on="s_info_windcode")
df.sort_values(by=[
"s_info_windcode", "report_period", "actual_ann_dt",
"statement_type"
],
inplace=True)
current_df = df.groupby(by="s_info_windcode").last()
data_series = [(dt.datetime.strftime(dt.datetime.now(), "%Y%m%d"),
list(current_df.index), current_df)]
for snapshot_date in report_period_generator(period=24, date=date):
df_slice = df[df["report_period"] <= snapshot_date].copy()
df_slice.sort_values(by=[
"s_info_windcode", "report_period", "actual_ann_dt",
"statement_type"
],
inplace=True)
df_slice = df_slice.groupby(by="s_info_windcode").last()
data_series.append([snapshot_date, list(df_slice.index), df_slice])
data_df = pd.DataFrame(data_series,
columns=["date", "ticker_list", "report"])
for ticker in data_df.loc[0, "ticker_list"]:
level = 0
while level < 20 and ticker in data_df.loc[level + 1,
"ticker_list"]:
level += 1
data_df.loc[0, "report"].loc[ticker, "max_level"] = level
data_df.loc[0, "report"]["max_level"] = data_df.loc[
0, "report"]["max_level"].astype(int)
print("Calculating TTM ...")
shift = Quarters2LastDec(date)
for ticker in data_df.loc[0, "ticker_list"]:
if data_df.loc[0, "report"].loc[ticker, "max_level"] >= 5:
for col in data_df.loc[0, "report"].columns[3:-1]:
data_df.loc[0, "report"].loc[ticker, col] = data_df.loc[0, "report"].loc[ticker, col] \
+ data_df.loc[shift, "report"].loc[ticker, col] - data_df.loc[5, "report"].loc[ticker, col]
print("Making snapshots ...")
df = data_df.loc[0, "report"].copy()
df.drop(
["report_period", "actual_ann_dt", "statement_type", "max_level"],
axis=1,
inplace=True)
for factor in df.columns:
df[factor + "_snapshots"] = [
dict(),
] * df.shape[0]
for col in df.columns:
if col.endswith("_snapshots"):
for index in df.index:
df.set_value(index, col, dict())
for ticker in data_df.loc[0, "ticker_list"]:
max_level = data_df.loc[0, "report"].loc[ticker, "max_level"]
for factor in data_df.loc[0, "report"].columns[
3:
-1]: # Exclude ["report_period", "actual_ann_dt", "statement_type", "max_level"]
for level in range(max_level, 0, -1):
df.loc[ticker, factor + "_snapshots"][data_df.loc[level, "date"]] = \
data_df.loc[level, "report"].loc[ticker, factor]
for col in df.columns:
if col.endswith("_snapshots"):
df[col] = df[col].astype(str)
print("Converting some NaN to 0 ...")
for col in df.columns:
if col.endswith("_snapshots"):
df[col] = df[col].astype(str)
for factor in [
"operating_cashflow", "operating_cashinflow",
"operating_cashoutflow", "investment_cashinflow",
"investment_cashoutflow", "investment_cashflow",
"finance_cashinflow", "finance_cashoutflow", "finance_cashflow"
]:
df[factor + "_snapshots"] = df[factor + "_snapshots"].apply(
lambda s: s.replace("{}", ''))
df[factor] = df[factor].apply(lambda x: 0 if pd.isna(x) else x)
df[factor + "_snapshots"] = df[factor + "_snapshots"].apply(
lambda s: s.replace("nan", '0'))
df = df.reset_index()
df.rename(columns={'s_info_windcode': 'code'}, inplace=True)
df["trade_date"] = date
new_cols = list(df.columns)
new_cols.insert(0, new_cols.pop(new_cols.index(
"trade_date"))) # Put trade_date into the first column
df = df[new_cols]
return df
