From 8f57eca962d8ece345e6e9810ab5ebe95731e9e8 Mon Sep 17 00:00:00 2001
From: Mathu-lmn <80094438+Mathu-lmn@users.noreply.github.com>
Date: Tue, 10 Dec 2024 22:00:46 -0500
Subject: [PATCH] =?UTF-8?q?alors=20peut=20=C3=AAtre=20=3F=3F=3F=3F?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 predictions.py | 16 ++++++++++------
 test.py        | 21 +++++++++++++++++++--
 2 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/predictions.py b/predictions.py
index 6e0e4a2..16db927 100644
--- a/predictions.py
+++ b/predictions.py
@@ -19,7 +19,7 @@
     from sklearn.ensemble import RandomForestClassifier
     from sklearn.neighbors import KNeighborsClassifier
     from sklearn.tree import DecisionTreeClassifier
-    from sklearn.metrics import accuracy_score, f1_score
+    from sklearn.metrics import classification_report, f1_score
     # from sklearn.model_selection import StratifiedKFold
 except ImportError:
     print("You need to install scikit-learn")
@@ -102,11 +102,14 @@
     model = models[name]
     param_dist = param_distributions[name]
     
+    def scoringfunc(estimator, X, y):
+        return f1_score(y, estimator.predict(X))
+    
     randomized_search = RandomizedSearchCV(
         estimator=model,
         param_distributions=param_dist,
         n_iter=NUM_TRIALS,
-        scoring='f1',
+        scoring=scoringfunc,
         cv=3,
         random_state=42,
         n_jobs=-1,
@@ -119,7 +122,9 @@
     model.set_params(**best)
     model.fit(X_train, y_train.values.ravel())
     
-    train_accuracy = accuracy_score(y_train, model.predict(X_train))
+    if DEBUG:
+        print(classification_report(y_test, model.predict(X_test)))
+
     f1_test = f1_score(y_test, model.predict(X_test))
     
     signature = infer_signature(X_train, model.predict(X_train))
@@ -127,10 +132,9 @@
     with mlflow.start_run(run_name=name) as run:
         mlflow.log_params(best)
         mlflow.sklearn.log_model(model, "model", signature=signature)
-        model_uri = mlflow.get_artifact_uri("model")
-        
+
         mlflow.evaluate(
-            model=model_uri,
+            model="runs:/" + run.info.run_id + "/model",
             data=evalData,
             targets='label',
             model_type='classifier',
diff --git a/test.py b/test.py
index 1217b81..bf00c20 100644
--- a/test.py
+++ b/test.py
@@ -4,9 +4,11 @@
 import random
 from sklearn.model_selection import train_test_split
 from sklearn.preprocessing import StandardScaler
+from sklearn.metrics import classification_report, f1_score
 
 mlflow.set_tracking_uri("https://mlflow.docsystem.xyz")
-model = mlflow.pyfunc.load_model(model_uri="models:/attackdetection/latest")
+# load each models in the RandomSearch experiment
+model = mlflow.sklearn.load_model("runs:/12498379aa574e01b6da4cff313479dd/model")
 
 LABELS_NUM = ["BENIGN", "SUS"]
 
@@ -27,6 +29,21 @@
 X_train = pd.DataFrame(X_train, columns=features.columns)
 X_test = pd.DataFrame(X_test, columns=features.columns)
 
+X_train = X_train.reset_index(drop=True)
+y_train = y_train.reset_index(drop=True)
+X_test = X_test.reset_index(drop=True)
+y_test = y_test.reset_index(drop=True)
+
+
+# calculate the model's metrics
+def calculate_metrics():
+    y_pred = model.predict(X_test)
+    print(classification_report(y_test, y_pred))
+
+    print(f1_score(y_test, y_pred))
+
+calculate_metrics()
+
 good_predictions, bad_predictions = 0, 0
 
 def predict_row(row: int):
@@ -41,7 +58,7 @@ def predict_row(row: int):
 
     return LABELS_NUM[y_test.iloc[row].values[0]] == prediction
 
-while True:
+while False:
     for i in tqdm(range(10000)):
         j = random.randint(0, len(X_test) - 1)
         is_good_prediction = predict_row(j)