v3.8.2

pure_ocean_breeze/__init__.py

@@ -2,8 +2,8 @@
 一个量化多因子研究的框架，包含数据、回测、因子加工等方面的功能
 """
 
-__updated__ = "2023-03-06 11:23:05"
-__version__ = "3.8.1"
+__updated__ = "2023-03-08 13:36:51"
+__version__ = "3.8.2"
 __author__ = "chenzongwei"
 __author_email__ = "[email protected]"
 __url__ = "https://github.com/chen-001/pure_ocean_breeze"

pure_ocean_breeze/data/database.py

@@ -1,4 +1,4 @@
-__updated__ = "2023-03-01 02:22:56"
+__updated__ = "2023-03-13 13:10:01"
 
 import pandas as pd
 import pymysql

pure_ocean_breeze/data/tools.py

@@ -2,7 +2,7 @@
 针对一些不常见的文件格式，读取数据文件的一些工具函数，以及其他数据工具
 """
 
-__updated__ = "2023-03-06 11:22:03"
+__updated__ = "2023-03-12 11:54:26"
 
 import os
 import pandas as pd
@@ -993,6 +993,28 @@ def get_list_std(delta_sts: list[pd.DataFrame]) -> pd.DataFrame:
     return delta_sts_std
 
 
+def get_list_std_weighted(delta_sts: list[pd.DataFrame], weights: list) -> pd.DataFrame:
+    """对多个df对应位置上的值求加权标准差
+
+    Parameters
+    ----------
+    delta_sts : list[pd.DataFrame]
+        多个dataframe
+    weights : list
+        权重序列
+
+    Returns
+    -------
+    pd.DataFrame
+        标准差序列
+    """    
+    weights = [i / sum(weights) for i in weights]
+    delta_sts_mean = sum(delta_sts) / len(delta_sts)
+    delta_sts_std = [(i - delta_sts_mean) ** 2 for i in delta_sts]
+    delta_sts_std = sum([i * j for i, j in zip(delta_sts_std, weights)])
+    return delta_sts_std**0.5
+
+
 @do_on_dfs
 def to_group(df: pd.DataFrame, group: int = 10) -> pd.DataFrame:
     """把一个index为时间，code为时间的df，每个截面上的值，按照排序分为group组，将值改为组号，从0开始
@@ -1215,7 +1237,10 @@ def get_fac_via_corr(
                         .fillna(method=fillna_method)
                         .dropna(axis=1)
                     )
-                corr = df0.corr(method=corr_method)
+                if corr_method == "spearman":
+                    corr = df0.rank().corr()
+                else:
+                    corr = df0.corr(method=corr_method)
                 if abs:
                     corr = corr.abs()
                 df0 = corr.mean().to_frame(end)
@@ -1254,7 +1279,10 @@ def get_fac_via_corr(
                     .fillna(method=fillna_method)
                     .dropna(axis=1)
                 )
-            corr = df0.corr(method=corr_method)
+            if corr_method == "spearman":
+                corr = df0.rank().corr()
+            else:
+                corr = df0.corr(method=corr_method)
             if abs:
                 corr = corr.abs()
             df0 = corr.mean().to_frame(end)

pure_ocean_breeze/data/write_data.py

@@ -1,4 +1,4 @@
-__updated__ = "2023-03-07 18:08:00"
+__updated__ = "2023-03-10 12:14:50"
 
 import time
 
@@ -51,6 +51,7 @@
 import pickledb
 import tqdm.auto
 from functools import reduce
+from typing import Union
 import dcube as dc
 from tenacity import retry, stop_after_attempt
 import questdb.ingress as qdbing
@@ -1108,9 +1109,7 @@ def database_update_industry_rets_for_stock():
     logger.success("股票对应申万一级行业每日收益率已经更新完")
 
 
-
-class FactorReader():
-
+class FactorReader:
     def __init__(
         self,
         user: str = "admin",
@@ -1139,12 +1138,7 @@ def __init__(
         self.host = host
         self.port = port
         self.database = database
-        infos=self.show_all_factors_information()
-        self.keys=list(infos.数据键名)
-        self.names=list(infos.因子名称)
-        self.keys_names = {k: v for k, v in zip(self.keys, self.names)}
-        self.names_keys = {k: v for k, v in zip(self.names, self.keys)}
-
+
     def __connect(self):
         conn = pg.connect(
             user=self.user,
@@ -1154,27 +1148,27 @@ def __connect(self):
             database=self.database,
         )
         return conn
-    
-    def __update_factor(self,table_name:str,df:pd.DataFrame):
-        tables=self.__get_data("show tables").table.tolist()
+
+    def update_factor(self, table_name: str, df: pd.DataFrame):
+        tables = self.__get_data("show tables").table.tolist()
         if table_name in tables:
-            logger.info(f'{table_name}已经存在了，即将更新')
-            old_end=self.__get_data(f"select max(date) from {table_name}").iloc[0,0]
-            new=df[df.index>old_end]
-            new=new.stack().reset_index()
-            new.columns=['date','code','fac']
+            logger.info(f"{table_name}已经存在了，即将更新")
+            old_end = self.__get_data(f"select max(date) from {table_name}").iloc[0, 0]
+            new = df[df.index > old_end]
+            new = new.stack().reset_index()
+            new.columns = ["date", "code", "fac"]
         else:
-            logger.info(f'{table_name}第一次上传')
-            new=df.stack().reset_index()
-            new.columns=['date','code','fac']
-        self.__write_via_df(new,table_name)
-        
+            logger.info(f"{table_name}第一次上传")
+            new = df.stack().reset_index()
+            new.columns = ["date", "code", "fac"]
+        self.__write_via_df(new, table_name)
+
     def __write_via_df(
         self,
         df: pd.DataFrame,
         table_name: str,
-        symbols= None,
-        tuple_col= None,
+        symbols=None,
+        tuple_col=None,
     ) -> None:
         """通过questdb的python库直接将dataframe写入quested数据库
 
@@ -1202,11 +1196,9 @@ def __write_via_df(
         else:
             with qdbing.Sender(self.host, 9009) as sender:
                 sender.dataframe(df, table_name=table_name)
-                
+
     @retry(stop=stop_after_attempt(10))
-    def __get_data(
-        self, sql_order: str
-    ) -> pd.DataFrame:
+    def __get_data(self, sql_order: str) -> pd.DataFrame:
         """以sql命令的方式，从数据库中读取数据
 
         Parameters
@@ -1225,4 +1217,8 @@ def __get_data(
         df_data = cursor.fetchall()
         columns = [i[0] for i in cursor.description]
         df = pd.DataFrame(df_data, columns=columns)
-        return df
+        return df
+
+    def add_token(self, tokens: list[str], users: list[str]):
+        tus = pd.DataFrame({"token": tokens, "user": users})
+        self.__write_via_df(tus, "tokenlines")

pure_ocean_breeze/labor/comment.py

@@ -1,4 +1,4 @@
-__updated__ = "2023-02-03 10:41:10"
+__updated__ = "2023-03-10 15:41:06"
 
 import numpy as np
 import pandas as pd
@@ -178,28 +178,35 @@ def make_relative_comments(
         如果没指定任何一个指数，将报错
     """
 
-    if hs300 and zz500:
+    if hs300 == 1 and zz500 == 1 and zz1000 == 0 and gz2000 == 0:
         net_index = read_index_single("000906.SH").resample("M").last()
     else:
         net_indexs = []
+        weights = []
         if hs300:
             net_index = read_index_single("000300.SH").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(300)
         if zz500:
             net_index = read_index_single("000905.SH").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(500)
         if zz1000:
             net_index = read_index_single("000852.SH").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(1000)
         if gz2000:
             net_index = read_index_single("399303.SZ").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(2000)
         if (hs300 + zz500 + zz1000 + gz2000) == 0:
             raise IOError("你总得指定一个股票池吧？")
         net_index = pd.concat(net_indexs, axis=1)
     ret_index = net_index.pct_change()
     if isinstance(ret_index, pd.DataFrame):
-        ret_index = ret_index.mean(axis=1)
+        ret_index = sum(
+            [ret_index.iloc[:, i] * weights[i] for i in range(len(weights))]
+        ) / sum(weights)
     if day is not None:
         ret_index = ret_index[ret_index.index >= pd.Timestamp(day)]
     ret = ret_fac - ret_index
@@ -252,28 +259,35 @@ def make_relative_comments_plot(
     `IOError`
         如果没指定任何一个指数，将报错
     """
-    if hs300 and zz500:
+    if hs300 == 1 and zz500 == 1 and zz1000 == 0 and gz2000 == 0:
         net_index = read_index_single("000906.SH").resample("M").last()
     else:
         net_indexs = []
+        weights = []
         if hs300:
             net_index = read_index_single("000300.SH").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(300)
         if zz500:
             net_index = read_index_single("000905.SH").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(500)
         if zz1000:
             net_index = read_index_single("000852.SH").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(1000)
         if gz2000:
             net_index = read_index_single("399303.SZ").resample("M").last()
             net_indexs.append(net_index)
+            weights.append(2000)
         if (hs300 + zz500 + zz1000 + gz2000) == 0:
             raise IOError("你总得指定一个股票池吧？")
         net_index = pd.concat(net_indexs, axis=1)
     ret_index = net_index.pct_change()
     if isinstance(ret_index, pd.DataFrame):
-        ret_index = ret_index.mean(axis=1)
+        ret_index = sum(
+            [ret_index.iloc[:, i] * weights[i] for i in range(len(weights))]
+        ) / sum(weights)
     if day is not None:
         ret_index = ret_index[ret_index.index >= pd.Timestamp(day)]
     ret = ret_fac - ret_index