stock_new更新

This commit is contained in:
muliangshu
2025-12-24 08:18:19 +08:00
parent 70224430b9
commit 21f6354658
+95
View File
@@ -0,0 +1,95 @@
import datetime
from sqlalchemy import create_engine
from dateutil.relativedelta import relativedelta
import pandas as pd
from sklearn.preprocessing import StandardScaler
from sklearn.mixture import GaussianMixture
import numpy as np
from collections import defaultdict
engine = create_engine('mysql+pymysql://root:Goojia123456@192.168.0.93:3306/stock')
mu_stocks_all = pd.read_sql(f"select * from mu_stocks_new", con = engine)
stock_basic_infos = pd.read_sql(f"select * from stock_basic", con=engine)
trade_dates = mu_stocks_all['trade_date'].tolist()
today = datetime.datetime.now()
def gmm_trade(get_data,df):
gmm = GaussianMixture(n_components=100, covariance_type='spherical', random_state=0)
data = df.values
data = np.array(data, dtype='float64')
x = data
gmm.fit(x)
y_hat = gmm.predict(x)
gmm_means_ = [i[0] for i in gmm.means_]
d = defaultdict(list)
for k, va in [(v, i) for i, v in enumerate(gmm_means_)]:
d[va] = k
gmm_means_list = []
for label in y_hat:
mean_value = d[label]
gmm_means_list.append(mean_value)
get_data["rating"] = y_hat
get_data["mean"] = gmm_means_list
return get_data
def stock_gmm(last_day_str,last_month_str):
start = datetime.datetime.now()
stock_all = pd.read_sql(f"select * from stock_index_daily where trade_date>'{last_month_str}' and trade_date<='{last_day_str}'", con = engine)
end = datetime.datetime.now()
print('MySQL Query Running time: %s Seconds' % (end - start))
stock_all = stock_all.sort_values(by="trade_date", ascending=False)
stock_select = stock_all[((stock_all['ts_code'].str.startswith('600')) | (stock_all['ts_code'].str.startswith('00')))]
stock_data = stock_select.dropna(subset=['turnover_rate_f', 'volume_ratio', 'pe', 'pe_ttm', 'pb', 'ps_ttm', 'ps'])
stock_group = stock_data.groupby(['ts_code'])
scaler = StandardScaler()
import_col = ['turnover_rate_f', 'volume_ratio', 'pe', 'pe_ttm', 'pb', 'ps_ttm', 'ps']
scaler_col = [str(i) for i in range(len(import_col))]
res = []
for ts_code, ts_detail in stock_group:
if len(ts_detail) > 3:
ts_detail[scaler_col] = ts_detail.apply(lambda x: (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0), axis=1,
result_type="expand")
ts_detail[scaler_col] = scaler.fit_transform(ts_detail[import_col].values)
for i, row in ts_detail.iterrows():
res.append(row)
stock_res = pd.DataFrame(res)
stock_all_newest = stock_all[stock_all.trade_date == stock_all.trade_date.max()]
stock_all_newest = stock_all_newest[['ts_code', 'turnover_rate', 'close', 'amount', 'pe', 'pe_ttm', 'total_mv', 'pct_chg']]
X = stock_res[['0', '1', '2', '3', '4', '5', '6']]
gmm_df = gmm_trade(stock_res, X)
gmm_group_df = gmm_df.groupby(['ts_code', 'rating'])['stock_name'].count().reset_index()
gmm_group_df.rename(columns={'stock_name': 'count'}, inplace=True)
gmm_df = gmm_df.merge(gmm_group_df, how='left', on=['ts_code', 'rating'])
min_mean = gmm_df['mean'].min()
slecct_df = gmm_df[gmm_df['mean'] == min_mean]
count_max = slecct_df['count'].max()
slecct_df = slecct_df[slecct_df['count'] == count_max]
slecct_df.drop_duplicates(subset=["ts_code"], keep="first", inplace=True)
block_list = list(slecct_df['stock_name'])
select_block = []
for name in block_list:
status = 1
for i in ['N', 'ST']:
if i in name:
status = 0
if status == 1:
select_block.append(name)
return stock_all_newest,select_block,slecct_df
def stock_select(i):
last_day = today+relativedelta(days=-i)
last_month = last_day+relativedelta(months=-2)
last_day_str = last_day.strftime('%Y%m%d')
if last_day_str in trade_dates:
return last_day_str
if int(last_day_str)<20250101:
return last_day_str
last_month_str = last_month.strftime('%Y%m%d')
stock_all_newest,select_block,slecct_df=stock_gmm(last_day_str,last_month_str)
db_stock = slecct_df[['ts_code', 'stock_name', 'open', 'close', 'high', 'low',
'vol', 'amount', 'pre_close', 'change', 'pct_chg', 'turnover_rate',
'turnover_rate_f', 'volume_ratio', 'pe', 'pe_ttm', 'pb', 'ps_ttm', 'ps',
'total_share', 'float_share', 'free_share', 'total_mv', 'circ_mv']].copy()
db_stock['trade_date'] = last_day_str
db_stock['ts_code'] = db_stock['ts_code'].apply(lambda x: x[0:6])
db_stock.to_sql("mu_stocks_new",con=engine, index=False, if_exists="append")
return last_day_str
if __name__ == '__main__':
stock_select(0)