param checking in treat_outliers

src/Dataset.py

@@ -160,7 +160,26 @@ class Dataset:
         is_competitive_col = self.get_is_competitive_col(dataframe)
         return is_competitive_col
 
-    def treat_outliers(self, df, colname) -> pd.DataFrame:
+    def treat_outliers(self, df: pd.DataFrame, colname: str) -> pd.DataFrame:
+        """Treat outliers of numerical columns.
+
+        Args:
+            df (pd.DataFrame): the dataframe.
+            colname (str): the column name to treat.
+
+        Returns:
+            pd.DataFrame: the filtered dataframe.
+        """
+        if type(df) != pd.DataFrame:
+            logging.error("parameter `dataframe` is not a pandas DataFrame")
+            raise ValueError(f"{df} is not a pandas DataFrame")
+        if type(colname) != str:
+            logging.error("parameter `colname` is not a string")
+            raise ValueError(f"{colname} is not a string")
+        if colname not in df.columns:
+            logging.error("column requested not in dataframe")
+            raise KeyError(f"{colname} is not a column in dataframe")
+
         q = df[colname].quantile(0.99)
         return df[df[colname] < q]