From c70cbc7141a768e7b7024e4577eb46f24e94a9f2 Mon Sep 17 00:00:00 2001
From: Franziska Oschmann
<franziskaoschmann@staff-net-oct-dock-1-a-dhcp-100.intern.ethz.ch>
Date: Thu, 13 Jul 2023 11:26:02 +0200
Subject: [PATCH] Fix issue with F1 computing
---
moderation_classifier/eval_MNB.py | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/moderation_classifier/eval_MNB.py b/moderation_classifier/eval_MNB.py
index 154ae18..7867c33 100644
--- a/moderation_classifier/eval_MNB.py
+++ b/moderation_classifier/eval_MNB.py
@@ -30,7 +30,7 @@ def main(train_logs: Union[str, os.PathLike]):
# Load logs
df = pd.read_csv(train_logs, index_col="Unnamed: 0")
- path_model = df.loc["path"].values[0]
+ path_model = df.loc["path_model"].values[0]
input_data = df.loc["input_data"].values[0].replace("train", "test")
# Load model
@@ -40,6 +40,7 @@ def main(train_logs: Union[str, os.PathLike]):
tl = TextLoader(input_data)
df_test = tl.load_text_csv(
newspaper="tagesanzeiger",
+ lang='de',
load_subset=False,
remove_duplicates=False,
min_num_words=3,
@@ -51,17 +52,16 @@ def main(train_logs: Union[str, os.PathLike]):
y_pred = pipe.predict(X_test)
y_pred_t = pipe.predict(X_test)
- precision, recall, *_ = precision_recall_fscore_support(
+ precision, recall, f1, _ = precision_recall_fscore_support(
y_test, y_pred, average="weighted"
)
- f1 = f1_score(y_test, y_pred)
- score = pipe.score(X_test, y_test)
+ accuracy = pipe.score(X_test, y_test)
results_all = dict()
results_all["precision"] = precision
results_all["recall"] = recall
results_all["f1"] = f1
- results_all["score"] = score
+ results_all["accuracy"] = accuracy
#import pdb; pdb.set_trace()
@@ -75,10 +75,10 @@ def main(train_logs: Union[str, os.PathLike]):
y_test_t = df_test[df_test.topic == t].label
y_pred_t = pipe.predict(X_test_t)
- precision, recall, *_ = precision_recall_fscore_support(
+ precision, recall, f1, _ = precision_recall_fscore_support(
y_test_t, y_pred_t, average="weighted"
)
- f1 = f1_score(y_test_t, y_pred_t)
+ #f1 = f1_score(y_test_t, y_pred_t)
accuracy = pipe.score(X_test_t, y_test_t)
results_t[t] = dict()
results_t[t]["accuracy"] = accuracy
--
GitLab