Merge pull request #11 from liamdugan/fix_evaluate

Fix `run_evaluation` bug

raid/_version.py

@@ -1 +1 @@
-__version__ = "0.0.6"
+__version__ = "0.0.7"

raid/detect.py

@@ -1,9 +1,12 @@
 def run_detection(f, df):
+    # Make a copy of the IDs of the original dataframe to avoid editing in place
+    scores_df = df[["id"]].copy()
+
     # Run the detector function on the dataset and put output in score column
-    df["score"] = f(df["generation"])
+    scores_df["score"] = f(df["generation"])
 
     # Convert scores and ids to dict in 'records' format for seralization
     # e.g. [{'id':'...', 'score':0}, {'id':'...', 'score':1}, ...]
-    results = df[["id", "score"]].to_dict(orient="records")
+    results = scores_df[["id", "score"]].to_dict(orient="records")
 
     return results

raid/evaluate.py

@@ -7,6 +7,10 @@ def load_detection_result(df, results):
     # Load the dataframe and read in the scores
     scores_df = pd.DataFrame.from_records(results)
 
+    # If df has a pre-existing score column, remove it before merging
+    if "score" in df.columns:
+        df = df.drop(columns=["score"])
+
     # Merge dataframes based on the id and validate that ids are unique
     return df.join(scores_df.set_index("id"), on="id", validate="one_to_one")