app.py

import gradio as gr
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split, cross_val_score, GridSearchCV
from sklearn.ensemble import RandomForestRegressor, RandomForestClassifier, GradientBoostingRegressor, GradientBoostingClassifier
from sklearn.linear_model import LinearRegression, LogisticRegression
from sklearn.metrics import r2_score, mean_absolute_error, mean_squared_error, accuracy_score, classification_report, confusion_matrix
from sklearn.preprocessing import LabelEncoder, StandardScaler
from sklearn.impute import SimpleImputer
import json
import bcrypt
import os
from datetime import datetime
import joblib
import warnings
warnings.filterwarnings('ignore')

# Authentication system with bcrypt
class AuthSystem:
    def __init__(self, auth_file="user_auth.json"):
        self.auth_file = auth_file
        self.load_auth_data()
    
    def load_auth_data(self):
        """Load authentication data from JSON file"""
        if os.path.exists(self.auth_file):
            with open(self.auth_file, 'r') as f:
                self.auth_data = json.load(f)
        else:
            # Initialize with default credentials
            self.auth_data = {
                "username": "admin",
                "password_hash": self.hash_password("admin123"),
                "created_at": datetime.now().isoformat()
            }
            self.save_auth_data()
    
    def save_auth_data(self):
        """Save authentication data to JSON file"""
        with open(self.auth_file, 'w') as f:
            json.dump(self.auth_data, f, indent=2)
    
    def hash_password(self, password):
        """Hash password using bcrypt"""
        return bcrypt.hashpw(password.encode(), bcrypt.gensalt()).decode()
    
    def verify_password(self, password, hashed):
        """Verify password against hash"""
        return bcrypt.checkpw(password.encode(), hashed.encode())
    
    def register(self, username, password):
        """Register new user (overwrites existing)"""
        if not username or not password:
            return False, "Kullanıcı adı ve şifre gereklidir"
        
        if len(password) < 6:
            return False, "Şifre en az 6 karakter olmalıdır"
        
        self.auth_data = {
            "username": username,
            "password_hash": self.hash_password(password),
            "created_at": datetime.now().isoformat()
        }
        self.save_auth_data()
        return True, "Kullanıcı başarıyla kaydedildi"
    
    def login(self, username, password):
        """Authenticate user"""
        if not username or not password:
            return False, "Kullanıcı adı ve şifre gereklidir"
        
        if (self.auth_data.get("username") == username and 
            self.verify_password(password, self.auth_data.get("password_hash"))):
            return True, "Giriş başarılı"
        else:
            return False, "Geçersiz kullanıcı bilgileri"

# Initialize authentication system
auth_system = AuthSystem()

def authenticate_user(username, password):
    """Authentication function for Gradio"""
    success, message = auth_system.login(username, password)
    return success, message

def register_user(username, password):
    """Registration function for Gradio"""
    success, message = auth_system.register(username, password)
    return success, message

def analyze_data(file, progress=gr.Progress()):
    """Enhanced data analysis function with progress tracking"""
    if file is None:
        fig, ax = plt.subplots(figsize=(1,1))
        fig.patch.set_visible(False)
        ax.axis('off')
        plt.close(fig)
        return "Lütfen bir CSV dosyası yükleyin.", fig, fig, None, None, None, [], None, None, None, None, None

    try:
        progress(0.1, desc="Dosya okunuyor...")
        
        # Check file size (limit to 100MB)
        file_size = os.path.getsize(file.name)
        if file_size > 100 * 1024 * 1024:
            raise Exception("Dosya boyutu 100MB'dan küçük olmalıdır")
        
        # Read CSV with error handling
        try:
            df = pd.read_csv(file.name, encoding='utf-8')
        except UnicodeDecodeError:
            df = pd.read_csv(file.name, encoding='latin-1')
        
        progress(0.3, desc="Veri analiz ediliyor...")
        
        # Basic dataset information
        num_rows, num_cols = df.shape
        column_info = "\n".join([f"- {col} ({df[col].dtype})" for col in df.columns])
        
        # Missing values analysis
        missing_values_info = df.isnull().sum()
        missing_values_report = "Eksik Değerler:\n" + missing_values_info[missing_values_info > 0].to_string()
        if missing_values_info.sum() == 0:
            missing_values_report = "Eksik değer bulunamadı."
        
        progress(0.5, desc="Sütunlar analiz ediliyor...")
        
        # Column explanations (automatic analysis)
        column_explanations = generate_column_explanations(df)
        
        progress(0.7, desc="Grafikler oluşturuluyor...")
        
        # Correlation matrix
        correlation_plot_figure = create_correlation_plot(df)
        
        # Feature importance plot (initially empty)
        feature_importance_fig = create_empty_plot()
        
        progress(0.9, desc="İstatistikler hesaplanıyor...")
        
        # Basic statistics
        numeric_df = df.select_dtypes(include=[np.number])
        basic_stats = ""
        if not numeric_df.empty:
            basic_stats = "Temel İstatistikler:\n" + numeric_df.describe().to_string()
        
        # Outlier detection
        outlier_info = detect_outliers(df)
        
        summary_text = (
            f"## ✅ Veri Seti Başarıyla Yüklendi!\n\n"
            f"- **Satır Sayısı:** {num_rows:,}\n"
            f"- **Sütun Sayısı:** {num_cols}\n"
            f"- **Dosya Boyutu:** {file_size / 1024:.2f} KB\n\n"
            f"### 📋 Sütun Bilgileri:\n```\n{column_info}\n```\n\n"
            f"### ⚠️ Eksik Değer Raporu:\n```\n{missing_values_report}\n```\n\n"
            f"### 🔍 Aykırı Değer Tespiti:\n```\n{outlier_info}\n```\n\n"
            f"### 📊 Temel İstatistikler (Sayısal Sütunlar):\n```\n{basic_stats}\n```\n\n"
            f"### 👀 İlk 5 Satır:\n```\n{df.head().to_string(max_rows=5, max_cols=10)}\n```"
        )
        
        progress(1.0, desc="Tamamlandı!")
        
        return (summary_text, correlation_plot_figure, feature_importance_fig, column_explanations,
                "", "", [str(col) for col in df.columns], df, None, None, None, None)
    
    except Exception as e:
        fig, ax = plt.subplots(figsize=(1,1))
        fig.patch.set_visible(False)
        ax.axis('off')
        plt.close(fig)
        return f"❌ Dosya işlenirken hata oluştu: {str(e)}", fig, fig, "", "", "", [], None, None, None, None, None

def detect_outliers(df):
    """Detect outliers using IQR method"""
    numeric_df = df.select_dtypes(include=[np.number])
    outlier_report = []
    
    for col in numeric_df.columns:
        Q1 = numeric_df[col].quantile(0.25)
        Q3 = numeric_df[col].quantile(0.75)
        IQR = Q3 - Q1
        lower_bound = Q1 - 1.5 * IQR
        upper_bound = Q3 + 1.5 * IQR
        
        outliers = numeric_df[(numeric_df[col] < lower_bound) | (numeric_df[col] > upper_bound)][col]
        
        if len(outliers) > 0:
            outlier_report.append(f"{col}: {len(outliers)} aykırı değer tespit edildi ({len(outliers)/len(df)*100:.2f}%)")
    
    if not outlier_report:
        return "Aykırı değer tespit edilmedi."
    
    return "\n".join(outlier_report)

def generate_column_explanations(df):
    """Generate automatic explanations for each column"""
    explanations = []
    
    for col in df.columns:
        col_data = df[col]
        dtype = col_data.dtype
        
        if pd.api.types.is_numeric_dtype(col_data):
            # Numeric column analysis
            unique_count = col_data.nunique()
            null_count = col_data.isnull().sum()
            mean_val = col_data.mean()
            std_val = col_data.std()
            min_val = col_data.min()
            max_val = col_data.max()
            
            explanation = f"**{col}** (Sayısal):\n"
            explanation += f"- Benzersiz değerler: {unique_count}\n"
            explanation += f"- Eksik değerler: {null_count} ({null_count/len(df)*100:.2f}%)\n"
            explanation += f"- Ortalama: {mean_val:.2f}\n"
            explanation += f"- Standart sapma: {std_val:.2f}\n"
            explanation += f"- Min: {min_val:.2f}, Max: {max_val:.2f}\n"
            
            if unique_count < 10:
                explanation += f"- Olası kategorik değerler: {sorted(col_data.unique())}\n"
            
        else:
            # Categorical column analysis
            unique_count = col_data.nunique()
            null_count = col_data.isnull().sum()
            most_common = col_data.value_counts().head(3)
            
            explanation = f"**{col}** (Kategorik):\n"
            explanation += f"- Benzersiz değerler: {unique_count}\n"
            explanation += f"- Eksik değerler: {null_count} ({null_count/len(df)*100:.2f}%)\n"
            explanation += f"- En yaygın değerler:\n"
            for val, count in most_common.items():
                explanation += f"  * {val}: {count} ({count/len(df)*100:.2f}%)\n"
        
        explanations.append(explanation)
    
    return "\n\n".join(explanations)

def create_correlation_plot(df):
    """Create correlation matrix plot"""
    numeric_df = df.select_dtypes(include=[np.number])
    
    if numeric_df.empty or len(numeric_df.columns) < 2:
        return create_empty_plot()
    
    fig, ax = plt.subplots(figsize=(12, 10))
    corr_matrix = numeric_df.corr()
    sns.heatmap(corr_matrix, annot=True, cmap='coolwarm', fmt=".2f", ax=ax, 
                cbar_kws={'label': 'Korelasyon Katsayısı'})
    ax.set_title("Sayısal Sütunların Korelasyon Matrisi", fontsize=16, fontweight='bold')
    plt.tight_layout()
    return fig

def create_empty_plot():
    """Create empty plot"""
    fig, ax = plt.subplots(figsize=(1,1))
    fig.patch.set_visible(False)
    ax.axis('off')
    plt.close(fig)
    return fig

def run_ml_analysis(df_state, target_column, missing_strategy, scale_features, use_cv, n_cv_folds, progress=gr.Progress()):
    """Run comprehensive machine learning analysis"""
    if df_state is None:
        return "❌ Lütfen önce bir CSV dosyası yükleyin.", "", None, None, None, None
    
    if not target_column:
        return "❌ Lütfen bir hedef değişken seçin.", "", None, None, None, None
    
    if target_column not in df_state.columns:
        return f"❌ Hata: '{target_column}' sütunu veri setinde bulunamadı.", "", None, None, None, None
    
    try:
        progress(0.1, desc="Veri hazırlanıyor...")
        df = df_state.copy()
        
        # Remove rows with missing target values
        if df[target_column].isnull().any():
            initial_rows = len(df)
            df.dropna(subset=[target_column], inplace=True)
            rows_after_drop = len(df)
            
            if rows_after_drop == 0:
                return ("❌ Hata: Hedef değişkendeki eksik değerler nedeniyle tüm satırlar silindi.", 
                        "", None, None, None, None)
        
        # Prepare data
        X = df.drop(columns=[target_column])
        y = df[target_column]
        
        progress(0.2, desc="Kategorik değişkenler işleniyor...")
        
        # Handle categorical variables
        categorical_columns = X.select_dtypes(include=['object']).columns
        label_encoders = {}
        for col in categorical_columns:
            if X[col].isnull().any():
                X[col] = X[col].fillna('Missing')
            le = LabelEncoder()
            X[col] = le.fit_transform(X[col].astype(str))
            label_encoders[col] = le
        
        # Handle missing values in features
        numeric_feature_cols = X.select_dtypes(include=[np.number]).columns
        if not numeric_feature_cols.empty and X[numeric_feature_cols].isnull().any().any():
            if missing_strategy == "Ortalama":
                imputer = SimpleImputer(strategy='mean')
            elif missing_strategy == "Medyan":
                imputer = SimpleImputer(strategy='median')
            else:  # "Mod"
                imputer = SimpleImputer(strategy='most_frequent')
            
            X[numeric_feature_cols] = imputer.fit_transform(X[numeric_feature_cols])
        
        if X.empty or len(y) == 0:
            return ("❌ Hata: Ön işleme sonrası veri kalmadı.", "", None, None, None, None)

        if len(y) < 2:
            return ("❌ Hata: ML analizi için yeterli örnek yok.", "", None, None, None, None)
        
        progress(0.3, desc="Problem tipi belirleniyor...")
        
        # Determine problem type
        if y.dtype == 'object' or (pd.api.types.is_numeric_dtype(y) and y.nunique() < 20):
            problem_type = 'classification'
            le_y = LabelEncoder()
            y = le_y.fit_transform(y.astype(str))
        else:
            problem_type = 'regression'
        
        # Feature scaling
        if scale_features:
            progress(0.4, desc="Özellikler ölçeklendiriliyor...")
            scaler = StandardScaler()
            X = pd.DataFrame(scaler.fit_transform(X), columns=X.columns)
        
        progress(0.5, desc="Veri bölünüyor...")
        
        # Split data
        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
        
        results_text = ""
        feature_importance_data = None
        predictions_plot = None
        confusion_matrix_plot = None
        cv_results_text = ""
        
        if problem_type == 'regression':
            progress(0.6, desc="Regresyon modelleri eğitiliyor...")
            
            # Linear Regression
            lr_model = LinearRegression()
            lr_model.fit(X_train, y_train)
            lr_pred = lr_model.predict(X_test)
            lr_r2 = r2_score(y_test, lr_pred)
            lr_mae = mean_absolute_error(y_test, lr_pred)
            lr_rmse = np.sqrt(mean_squared_error(y_test, lr_pred))
            
            # Random Forest
            rf_model = RandomForestRegressor(n_estimators=100, random_state=42, n_jobs=-1)
            rf_model.fit(X_train, y_train)
            rf_pred = rf_model.predict(X_test)
            rf_r2 = r2_score(y_test, rf_pred)
            rf_mae = mean_absolute_error(y_test, rf_pred)
            rf_rmse = np.sqrt(mean_squared_error(y_test, rf_pred))
            
            # Gradient Boosting
            gb_model = GradientBoostingRegressor(n_estimators=100, random_state=42)
            gb_model.fit(X_train, y_train)
            gb_pred = gb_model.predict(X_test)
            gb_r2 = r2_score(y_test, gb_pred)
            gb_mae = mean_absolute_error(y_test, gb_pred)
            gb_rmse = np.sqrt(mean_squared_error(y_test, gb_pred))
            
            progress(0.8, desc="Sonuçlar hazırlanıyor...")
            
            # Feature importance
            feature_importance_data = pd.DataFrame({
                'feature': X.columns,
                'importance': rf_model.feature_importances_
            }).sort_values('importance', ascending=False)
            
            # Cross-validation
            if use_cv:
                progress(0.85, desc="Cross-validation yapılıyor...")
                rf_cv_scores = cross_val_score(rf_model, X, y, cv=n_cv_folds, scoring='r2')
                cv_results_text = f"\n\n🔄 Cross-Validation Sonuçları (Random Forest, {n_cv_folds}-Fold):\n"
                cv_results_text += f"- Ortalama R² Skoru: {rf_cv_scores.mean():.4f} (±{rf_cv_scores.std():.4f})\n"
                cv_results_text += f"- Min: {rf_cv_scores.min():.4f}, Max: {rf_cv_scores.max():.4f}"
            
            results_text = f"""📊 **Makine Öğrenimi Analiz Sonuçları (Regresyon)**

🔹 **Doğrusal Regresyon:**
- R² Skoru: {lr_r2:.4f}
- MAE: {lr_mae:.4f}
- RMSE: {lr_rmse:.4f}

🔹 **Rastgele Orman:**
- R² Skoru: {rf_r2:.4f}
- MAE: {rf_mae:.4f}
- RMSE: {rf_rmse:.4f}

🔹 **Gradient Boosting:**
- R² Skoru: {gb_r2:.4f}
- MAE: {gb_mae:.4f}
- RMSE: {gb_rmse:.4f}

🏆 **En İyi Model:** {"Gradient Boosting" if gb_r2 == max(lr_r2, rf_r2, gb_r2) else "Rastgele Orman" if rf_r2 == max(lr_r2, rf_r2, gb_r2) else "Doğrusal Regresyon"}

📈 **Özellik Önemi (İlk 10):**
{feature_importance_data.head(10).to_string(index=False)}

{cv_results_text}"""
            
            # Predictions plot
            predictions_plot = plot_predictions(y_test, rf_pred, "Rastgele Orman")
            
            # Save best model
            best_model = rf_model if rf_r2 >= max(lr_r2, gb_r2) else (gb_model if gb_r2 >= lr_r2 else lr_model)
            model_path = "saved_model.pkl"
            joblib.dump(best_model, model_path)
            
        else:  # Classification
            progress(0.6, desc="Sınıflandırma modelleri eğitiliyor...")
            
            # Logistic Regression
            lr_model = LogisticRegression(random_state=42, max_iter=1000, n_jobs=-1)
            lr_model.fit(X_train, y_train)
            lr_pred = lr_model.predict(X_test)
            lr_accuracy = accuracy_score(y_test, lr_pred)
            
            # Random Forest
            rf_model = RandomForestClassifier(n_estimators=100, random_state=42, n_jobs=-1)
            rf_model.fit(X_train, y_train)
            rf_pred = rf_model.predict(X_test)
            rf_accuracy = accuracy_score(y_test, rf_pred)
            
            # Gradient Boosting
            gb_model = GradientBoostingClassifier(n_estimators=100, random_state=42)
            gb_model.fit(X_train, y_train)
            gb_pred = gb_model.predict(X_test)
            gb_accuracy = accuracy_score(y_test, gb_pred)
            
            progress(0.8, desc="Sonuçlar hazırlanıyor...")
            
            # Feature importance
            feature_importance_data = pd.DataFrame({
                'feature': X.columns,
                'importance': rf_model.feature_importances_
            }).sort_values('importance', ascending=False)
            
            # Confusion matrix
            confusion_matrix_plot = plot_confusion_matrix(y_test, rf_pred, le_y.classes_)
            
            # Cross-validation
            if use_cv:
                progress(0.85, desc="Cross-validation yapılıyor...")
                rf_cv_scores = cross_val_score(rf_model, X, y, cv=n_cv_folds, scoring='accuracy')
                cv_results_text = f"\n\n🔄 Cross-Validation Sonuçları (Random Forest, {n_cv_folds}-Fold):\n"
                cv_results_text += f"- Ortalama Doğruluk: {rf_cv_scores.mean():.4f} (±{rf_cv_scores.std():.4f})\n"
                cv_results_text += f"- Min: {rf_cv_scores.min():.4f}, Max: {rf_cv_scores.max():.4f}"
            
            results_text = f"""📊 **Makine Öğrenimi Analiz Sonuçları (Sınıflandırma)**

🔹 **Lojistik Regresyon:**
- Doğruluk: {lr_accuracy:.4f}

🔹 **Rastgele Orman:**
- Doğruluk: {rf_accuracy:.4f}

🔹 **Gradient Boosting:**
- Doğruluk: {gb_accuracy:.4f}

🏆 **En İyi Model:** {"Gradient Boosting" if gb_accuracy == max(lr_accuracy, rf_accuracy, gb_accuracy) else "Rastgele Orman" if rf_accuracy == max(lr_accuracy, rf_accuracy, gb_accuracy) else "Lojistik Regresyon"}

📋 **Sınıflandırma Raporu (Rastgele Orman):**
{classification_report(y_test, rf_pred, target_names=le_y.classes_)}

📈 **Özellik Önemi (İlk 10):**
{feature_importance_data.head(10).to_string(index=False)}

{cv_results_text}"""
            
            # Save best model
            best_model = rf_model if rf_accuracy >= max(lr_accuracy, gb_accuracy) else (gb_model if gb_accuracy >= lr_accuracy else lr_model)
            model_path = "saved_model.pkl"
            joblib.dump(best_model, model_path)
        
        # Model summary
        model_summary = f"""🔧 **Model Bilgileri:**

- Problem Tipi: {problem_type.title()}
- Kullanılan Özellikler: {len(X.columns)}
- Eğitim Örnekleri: {len(X_train):,}
- Test Örnekleri: {len(X_test):,}
- Özellik Ölçeklendirme: {'Evet' if scale_features else 'Hayır'}
- Eksik Değer Stratejisi: {missing_strategy}
- Model Kaydedildi: saved_model.pkl"""
        
        # Feature importance plot
        feature_importance_fig = plot_feature_importance(feature_importance_data)
        
        progress(1.0, desc="Tamamlandı!")
        
        return (results_text, model_summary, feature_importance_fig, 
                predictions_plot if problem_type == 'regression' else confusion_matrix_plot,
                model_path, feature_importance_data)
    
    except Exception as e:
        return f"❌ ML analizi sırasında hata oluştu: {str(e)}", f"❌ Hata: {str(e)}", None, None, None, None

def plot_feature_importance(feature_importance_data):
    """Plot feature importance"""
    if feature_importance_data is None or feature_importance_data.empty:
        return create_empty_plot()
    
    fig, ax = plt.subplots(figsize=(10, 8))
    top_features = feature_importance_data.head(15)
    ax.barh(top_features['feature'], top_features['importance'], color='steelblue')
    ax.set_xlabel('Önem Skoru', fontsize=12)
    ax.set_ylabel('Özellik', fontsize=12)
    ax.set_title('En Önemli 15 Özellik', fontsize=14, fontweight='bold')
    ax.invert_yaxis()
    plt.tight_layout()
    return fig

def plot_predictions(y_test, y_pred, model_name):
    """Plot actual vs predicted values"""
    fig, ax = plt.subplots(figsize=(10, 8))
    ax.scatter(y_test, y_pred, alpha=0.6, edgecolors='k', s=50)
    
    # Perfect prediction line
    min_val = min(y_test.min(), y_pred.min())
    max_val = max(y_test.max(), y_pred.max())
    ax.plot([min_val, max_val], [min_val, max_val], 'r--', lw=2, label='Mükemmel Tahmin')
    
    ax.set_xlabel('Gerçek Değerler', fontsize=12)
    ax.set_ylabel('Tahmin Edilen Değerler', fontsize=12)
    ax.set_title(f'Gerçek vs Tahmin Edilen Değerler ({model_name})', fontsize=14, fontweight='bold')
    ax.legend()
    ax.grid(True, alpha=0.3)
    plt.tight_layout()
    return fig

def plot_confusion_matrix(y_test, y_pred, class_names):
    """Plot confusion matrix"""
    cm = confusion_matrix(y_test, y_pred)
    
    fig, ax = plt.subplots(figsize=(10, 8))
    sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', xticklabels=class_names, 
                yticklabels=class_names, ax=ax, cbar_kws={'label': 'Sayı'})
    ax.set_xlabel('Tahmin Edilen Sınıf', fontsize=12)
    ax.set_ylabel('Gerçek Sınıf', fontsize=12)
    ax.set_title('Confusion Matrix (Rastgele Orman)', fontsize=14, fontweight='bold')
    plt.tight_layout()
    return fig

def download_model_file(model_path):
    """Return model file for download"""
    if model_path and os.path.exists(model_path):
        return model_path
    return None

def export_results_to_csv(feature_importance_data):
    """Export feature importance to CSV"""
    if feature_importance_data is not None:
        output_path = "feature_importance.csv"
        feature_importance_data.to_csv(output_path, index=False)
        return output_path
    return None

# Create Gradio interface
def create_interface():
    with gr.Blocks(title="🤖 Gelişmiş Veri Analizi Sistemi", theme=gr.themes.Soft()) as demo:
        # Authentication state
        auth_state = gr.State(value=False)
        
        # State variables
        df_state = gr.State(value=None)
        saved_model_path_state = gr.State(value=None)
        feature_importance_state = gr.State(value=None)
        
        # Login Interface (always visible at start)
        with gr.Column(visible=True) as login_page:
            gr.Markdown(
                """
                # 🔐 Giriş Yap
                
                Sisteme erişmek için lütfen giriş yapın.
                """
            )
            
            with gr.Row():
                with gr.Column(scale=1):
                    pass
                with gr.Column(scale=1):
                    login_username = gr.Textbox(
                        label="Kullanıcı Adı", 
                        placeholder="Kullanıcı adınızı girin",
                        elem_id="login_username"
                    )
                    login_password = gr.Textbox(
                        label="Şifre", 
                        type="password", 
                        placeholder="Şifrenizi girin",
                        elem_id="login_password"
                    )
                    login_status = gr.Markdown("")
                    
                    with gr.Row():
                        login_button = gr.Button("🚀 Giriş Yap", variant="primary", size="lg", scale=2)
                        register_page_button = gr.Button("📝 Kayıt Ol", variant="secondary", size="lg", scale=1)
                    
                    gr.Markdown("---")
                    gr.Markdown("**💡 Varsayılan Giriş:** `admin` / `admin123`")
                with gr.Column(scale=1):
                    pass
        
        # Register Interface (hidden at start)
        with gr.Column(visible=False) as register_page:
            gr.Markdown(
                """
                # ✍️ Yeni Hesap Oluştur
                
                Sisteme kayıt olmak için bilgilerinizi girin.
                """
            )
            
            with gr.Row():
                with gr.Column(scale=1):
                    pass
                with gr.Column(scale=1):
                    register_username = gr.Textbox(
                        label="Kullanıcı Adı", 
                        placeholder="Yeni kullanıcı adı seçin"
                    )
                    register_password = gr.Textbox(
                        label="Şifre", 
                        type="password", 
                        placeholder="Şifre (min 6 karakter)"
                    )
                    register_confirm_password = gr.Textbox(
                        label="Şifre Tekrar", 
                        type="password", 
                        placeholder="Şifreyi tekrar girin"
                    )
                    register_status = gr.Markdown("")
                    
                    with gr.Row():
                        register_button = gr.Button("✅ Kayıt Ol", variant="primary", size="lg", scale=2)
                        back_to_login_button = gr.Button("◀️ Geri Dön", variant="secondary", size="lg", scale=1)
                with gr.Column(scale=1):
                    pass
        
        # Main Analysis Interface (hidden at start)
        with gr.Column(visible=False) as analysis_page:
            gr.Markdown("# 📊 Veri Analizi Sistemi")
            
            with gr.Row():
                with gr.Column(scale=1):
                    gr.Markdown("### 📂 Veri Seti Yükleme")
                    file_input = gr.File(label="CSV Dosyası Yükle (Max 100MB)", file_types=[".csv"])
                    run_button = gr.Button("🚀 Analizi Başlat", variant="primary", size="lg")
                    
                    gr.Markdown("---")
                    gr.Markdown("### 🤖 Makine Öğrenimi Ayarları")
                    
                    target_column_dropdown = gr.Dropdown(
                        label="🎯 Hedef Değişkeni Seç", 
                        choices=[], 
                        interactive=True,
                        info="Tahmin edilecek sütunu seçin"
                    )
                    
                    with gr.Accordion("⚙️ Gelişmiş Ayarlar", open=False):
                        missing_strategy = gr.Radio(
                            label="Eksik Değer Stratejisi",
                            choices=["Ortalama", "Medyan", "Mod"],
                            value="Ortalama",
                            info="Eksik değerleri nasıl doldurmak istersiniz?"
                        )
                        
                        scale_features = gr.Checkbox(
                            label="Özellik Ölçeklendirme (StandardScaler)",
                            value=True,
                            info="Özellikleri normalize et"
                        )
                        
                        use_cv = gr.Checkbox(
                            label="Cross-Validation Kullan",
                            value=True,
                            info="Model performansını daha iyi değerlendirmek için"
                        )
                        
                        n_cv_folds = gr.Slider(
                            label="CV Fold Sayısı",
                            minimum=3,
                            maximum=10,
                            value=5,
                            step=1,
                            info="Cross-validation için fold sayısı"
                        )
                    
                    run_ml_button = gr.Button("🎯 ML Analizi Çalıştır", variant="secondary", size="lg")
                    
                    gr.Markdown("---")
                    gr.Markdown("### 💾 İndirme Seçenekleri")
                    download_model_button = gr.Button("📦 Modeli İndir (.pkl)", size="sm")
                    model_download_file = gr.File(label="Model Dosyası", visible=False)
                    
                    download_feature_importance_button = gr.Button("📊 Özellik Önemini İndir (.csv)", size="sm")
                    feature_importance_download_file = gr.File(label="Özellik Önemi CSV", visible=False)
                
                with gr.Column(scale=2):
                    gr.Markdown("### 📈 Analiz Sonuçları")
                    
                    with gr.Tabs():
                        with gr.TabItem("📋 Veri Seti Genel Bakış"):
                            output_text = gr.Markdown(label="Analiz Sonuçları")
                        
                        with gr.TabItem("🔍 Sütun Açıklamaları"):
                            column_explanations_output = gr.Textbox(
                                label="Sütun Detaylı Açıklamaları", 
                                lines=20, 
                                interactive=False,
                                show_copy_button=True
                            )
                        
                        with gr.TabItem("🌡️ Korelasyon Matrisi"):
                            correlation_plot_output = gr.Plot(label="Sayısal Sütunların Korelasyon Matrisi")
                        
                        with gr.TabItem("🎯 ML Sonuçları"):
                            ml_results_output = gr.Textbox(
                                label="Makine Öğrenimi Modeli Sonuçları", 
                                lines=25, 
                                interactive=False,
                                show_copy_button=True
                            )
                            ml_model_summary = gr.Textbox(
                                label="Model Özeti", 
                                lines=10, 
                                interactive=False,
                                show_copy_button=True
                            )
                        
                        with gr.TabItem("📊 Özellik Önemi"):
                            feature_importance_plot = gr.Plot(label="En Önemli Özellikler")
                        
                        with gr.TabItem("📈 Tahmin Grafiği"):
                            predictions_plot = gr.Plot(label="Gerçek vs Tahmin / Confusion Matrix")
        
        # Page navigation functions
        def show_register_page():
            return gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)
        
        def show_login_page():
            return gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)
        
        def show_analysis_page():
            return gr.update(visible=False), gr.update(visible=False), gr.update(visible=True)
        
        # Authentication event handlers
        def handle_login(username, password):
            success, message = authenticate_user(username, password)
            if success:
                return (
                    True, 
                    f"✅ {message}", 
                    gr.update(visible=False),  # login_page
                    gr.update(visible=False),  # register_page
                    gr.update(visible=True)    # analysis_page
                )
            else:
                return (
                    False, 
                    f"❌ {message}", 
                    gr.update(visible=True),   # login_page
                    gr.update(visible=False),  # register_page
                    gr.update(visible=False)   # analysis_page
                )
        
        def handle_register(username, password, confirm_password):
            if password != confirm_password:
                return (
                    False, 
                    "❌ Şifreler eşleşmiyor!",
                    gr.update(visible=False),  # login_page
                    gr.update(visible=True),   # register_page
                    gr.update(visible=False)   # analysis_page
                )
            
            success, message = register_user(username, password)
            if success:
                return (
                    True, 
                    f"✅ {message} Giriş sayfasına yönlendiriliyorsunuz...",
                    gr.update(visible=True),   # login_page
                    gr.update(visible=False),  # register_page
                    gr.update(visible=False)   # analysis_page
                )
            else:
                return (
                    False, 
                    f"❌ {message}",
                    gr.update(visible=False),  # login_page
                    gr.update(visible=True),   # register_page
                    gr.update(visible=False)   # analysis_page
                )
        
        # Navigation button events
        register_page_button.click(
            show_register_page,
            outputs=[login_page, register_page, analysis_page]
        )
        
        back_to_login_button.click(
            show_login_page,
            outputs=[login_page, register_page, analysis_page]
        )
        
        # Authentication events
        login_button.click(
            handle_login,
            inputs=[login_username, login_password],
            outputs=[auth_state, login_status, login_page, register_page, analysis_page]
        )
        
        register_button.click(
            handle_register,
            inputs=[register_username, register_password, register_confirm_password],
            outputs=[auth_state, register_status, login_page, register_page, analysis_page]
        )
        
        # Analysis event handlers
        def update_dropdown_choices(df_state):
            """Update dropdown choices when new data is loaded"""
            if df_state is not None:
                valid_columns = [str(col) for col in df_state.columns if str(col).strip()]
                return gr.update(choices=valid_columns, value=None)
            return gr.update(choices=[], value=None)
        
        # Data loading
        run_button.click(
            analyze_data,
            inputs=file_input,
            outputs=[
                output_text, 
                correlation_plot_output, 
                feature_importance_plot,
                column_explanations_output, 
                ml_results_output, 
                ml_model_summary, 
                target_column_dropdown, 
                df_state, 
                saved_model_path_state,
                predictions_plot,
                model_download_file,
                feature_importance_download_file
            ]
        )
        
        # Update dropdown when data changes
        df_state.change(
            update_dropdown_choices,
            inputs=df_state,
            outputs=target_column_dropdown
        )
        
        # ML Analysis
        run_ml_button.click(
            run_ml_analysis,
            inputs=[
                df_state, 
                target_column_dropdown, 
                missing_strategy, 
                scale_features, 
                use_cv, 
                n_cv_folds
            ],
            outputs=[
                ml_results_output, 
                ml_model_summary, 
                feature_importance_plot,
                predictions_plot, 
                saved_model_path_state,
                feature_importance_state
            ]
        )
        
        # Download model
        download_model_button.click(
            download_model_file,
            inputs=saved_model_path_state,
            outputs=model_download_file
        )
        
        # Download feature importance
        download_feature_importance_button.click(
            export_results_to_csv,
            inputs=feature_importance_state,
            outputs=feature_importance_download_file
        )
    
    return demo

if __name__ == "__main__":
    demo = create_interface()
    demo.launch(
        share=False, 
        server_name="0.0.0.0",
       
        show_error=True,
        quiet=False
    )