feat: filter feature which is high correlation to former implemented features (#145)

* filter feature which is high correlation to former implemented features

* use multiprocessing to calculate IC and some minor fix

rdagent/components/coder/factor_coder/CoSTEER/evaluators.py

@@ -611,7 +611,6 @@ class FactorEvaluatorForCoder(FactorEvaluator):
                     value_feedback=factor_feedback.factor_value_feedback,
                     code_feedback=factor_feedback.code_feedback,
                 )
-            logger.info(factor_feedback.final_decision)
             return factor_feedback
 
 

rdagent/components/coder/factor_coder/config.py

@@ -44,7 +44,7 @@ class FactorImplementSettings(BaseSettings):
     max_loop: int = 10
 
     knowledge_base_path: Union[str, None] = None
-    new_knowledge_base_path: Union[str, None] = None
+    new_knowledge_base_path: Union[str, None] = knowledge_base_path
 
     python_bin: str = "python"
 

rdagent/components/coder/factor_coder/factor.py

@@ -70,7 +70,6 @@ class FactorFBWorkspace(FBWorkspace):
     ) -> None:
         super().__init__(*args, **kwargs)
         self.executed_factor_value_dataframe = executed_factor_value_dataframe
-        self.logger = logger
         self.raise_exception = raise_exception
 
     @staticmethod

rdagent/components/coder/model_coder/conf.py

@@ -15,7 +15,7 @@ class ModelImplSettings(BaseSettings):
     )
 
     knowledge_base_path: Union[str, None] = None
-    new_knowledge_base_path: Union[str, None] = None
+    new_knowledge_base_path: Union[str, None] = knowledge_base_path
 
     max_loop: int = 10
 

rdagent/components/knowledge_management/vector_base.py

@@ -127,8 +127,6 @@ class PDVectorBase(VectorBase):
         else:
             self.vector_df = pd.DataFrame(columns=["id", "label", "content", "embedding"])
 
-        logger.info(f"VectorBase loaded, shape={self.vector_df.shape}")
-
     def shape(self):
         return self.vector_df.shape
 
@@ -205,4 +203,3 @@ class PDVectorBase(VectorBase):
 
     def save(self, vector_df_path, **kwargs):
         self.vector_df.to_pickle(vector_df_path)
-        logger.info(f"Save vectorBase vector_df to: {vector_df_path}")