Create train_spam_model.py

train_spam_model.py

@@ -0,0 +1,96 @@
+# -----------------------------
+# 1️⃣ Import des librairies
+# -----------------------------
+import pandas as pd
+import re
+import joblib
+import nltk
+from nltk.corpus import stopwords
+from nltk.stem import PorterStemmer
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.model_selection import train_test_split
+from sklearn.linear_model import LogisticRegression
+from sklearn.metrics import classification_report, confusion_matrix, accuracy_score, roc_auc_score
+from imblearn.over_sampling import SMOTE
+
+# Télécharger stopwords si nécessaire
+nltk.download('stopwords')
+
+# -----------------------------
+# 2️⃣ Prétraitement des messages
+# -----------------------------
+stop_words = set(stopwords.words('english'))
+stemmer = PorterStemmer()
+
+def preprocess_message(text):
+    if pd.isna(text):
+        return ""
+    text = text.lower()
+    text = re.sub(r'http\S+|www\S+', '', text)  # supprimer URLs
+    text = re.sub(r'\S+@\S+', '', text)         # supprimer emails
+    text = re.sub(r'\+?\d[\d -]{8,}\d', '', text)  # supprimer numéros
+    text = re.sub(r'\d+', '', text)                # supprimer chiffres
+    text = re.sub(r'[^a-z\s!/+>]', '', text)      # garder ponctuation utile spam
+    words = [stemmer.stem(word) for word in text.split() if word not in stop_words]
+    return " ".join(words)
+
+# -----------------------------
+# 3️⃣ Charger les données
+# -----------------------------
+# data doit avoir les colonnes "Message" et "Category" ('spam'/'ham')
+data = pd.read_csv("data.csv")
+data['cleaned'] = data['Message'].apply(preprocess_message)
+
+X = data['cleaned']
+y = data['Category']
+
+# Split train/test stratifié
+X_train, X_test, y_train, y_test = train_test_split(
+    X, y, test_size=0.2, random_state=42, stratify=y
+)
+
+# -----------------------------
+# 4️⃣ Vectorisation TF-IDF
+# -----------------------------
+tfidf = TfidfVectorizer(
+    max_features=5000,
+    min_df=2,
+    max_df=0.95,
+    ngram_range=(1,2),
+    token_pattern=r'(?u)\b\w+\b|[!/+>]'  # capture mots et ponctuations importantes
+)
+X_train_tfidf = tfidf.fit_transform(X_train)
+X_test_tfidf = tfidf.transform(X_test)
+
+# -----------------------------
+# 5️⃣ Équilibrage des classes avec SMOTE
+# -----------------------------
+smote = SMOTE(random_state=42)
+X_train_balanced, y_train_balanced = smote.fit_resample(X_train_tfidf, y_train)
+
+# -----------------------------
+# 6️⃣ Entraînement du modèle Logistic Regression
+# -----------------------------
+model = LogisticRegression(random_state=42, max_iter=1000)
+model.fit(X_train_balanced, y_train_balanced)
+
+# -----------------------------
+# 7️⃣ Évaluation rapide
+# -----------------------------
+y_pred = model.predict(X_test_tfidf)
+print("Classification Report:\n", classification_report(y_test, y_pred))
+print("Matrice de confusion:\n", confusion_matrix(y_test, y_pred))
+accuracy = accuracy_score(y_test, y_pred)
+print(f"Accuracy: {accuracy:.4f}")
+
+if hasattr(model, 'predict_proba'):
+    y_test_binary = (y_test == 'spam').astype(int)
+    auc = roc_auc_score(y_test_binary, model.predict_proba(X_test_tfidf)[:,1])
+    print(f"AUC-ROC: {auc:.4f}")
+
+# -----------------------------
+# 8️⃣ Sauvegarder modèle et TF-IDF
+# -----------------------------
+joblib.dump(model, "spam_model.pkl")
+joblib.dump(tfidf, "tfidf_vectorizer.pkl")
+print("✅ Modèle Logistic Regression et TF-IDF vectorizer sauvegardés avec succès !")