train_minimal.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
train_minimal.py
----------------
Sadece subject, text, label kolonlarıyla minimal eğitim.
Kullanım:
python train_minimal.py --csv emails_dataset.csv
"""

import argparse
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import Pipeline
from sklearn.metrics import classification_report, roc_auc_score
import joblib

def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--csv", required=True, help="CSV yolu (kolonlar: subject,text,label)")
    ap.add_argument("--model", default="spam_model_min.joblib", help="Kaydedilecek model dosyası")
    ap.add_argument("--test-size", type=float, default=0.2, help="Test oranı")
    ap.add_argument("--seed", type=int, default=42, help="Rastgelelik tohumu")
    args = ap.parse_args()

    # Veri oku
    df = pd.read_csv(args.csv)
    for c in ["subject", "text", "label"]:
        if c not in df.columns:
            raise SystemExit(f"Kolon eksik: {c}. CSV kolonları: {list(df.columns)}")

    # Subject + text birleştir
    X = (df["subject"].fillna("").astype(str) + " \n " + df["text"].fillna("").astype(str))
    y = df["label"].astype(int)

    # Eğitim/Test böl
    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=args.test_size, random_state=args.seed, stratify=y
    )

    # Pipeline
    pipe = Pipeline([
        ("tfidf", TfidfVectorizer(
            lowercase=True,
            strip_accents="unicode",
            stop_words=None,        
            ngram_range=(1,2),
            min_df=1,               
            max_df=0.95
        )),
        ("clf", LogisticRegression(
            max_iter=2000,
            C=4.0,
            class_weight="balanced"
        ))
    ])

   
    pipe.fit(X_train, y_train)

    
    y_pred = pipe.predict(X_test)
    print(classification_report(y_test, y_pred, digits=3))

    if hasattr(pipe.named_steps["clf"], "predict_proba"):
        y_proba = pipe.predict_proba(X_test)[:,1]
        print("ROC-AUC:", roc_auc_score(y_test, y_proba))

   
    joblib.dump(pipe, args.model)
    print(f"Model kaydedildi: {args.model}")

if __name__ == "__main__":
    main()