create_demo_data.py

# create_demo_data.py - Cria dados de demonstração e modelos base
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, roc_auc_score
import joblib

def create_demo_dataset():
    """Cria dataset de demonstração realístico"""
    np.random.seed(42)
    n_samples = 2000
    
    # Features baseadas no dataset real de hotéis
    features = {
        'lead_time': np.random.gamma(2, 50, n_samples),
        'adr': np.random.normal(100, 30, n_samples),
        'adults': np.random.poisson(2, n_samples),
        'children': np.random.poisson(0.3, n_samples),
        'previous_cancellations': np.random.poisson(0.1, n_samples),
        'is_repeated_guest': np.random.binomial(1, 0.1, n_samples),
        'required_car_parking_spaces': np.random.binomial(1, 0.2, n_samples),
        'total_of_special_requests': np.random.poisson(0.5, n_samples),
        'booking_changes': np.random.poisson(0.3, n_samples),
    }
    
    X = pd.DataFrame(features)
    
    # Criar target com relação realística
    cancellation_prob = 1 / (1 + np.exp(-(
        X['lead_time'] * 0.01 +
        X['adr'] * 0.005 -
        X['is_repeated_guest'] * 0.8 -
        X['required_car_parking_spaces'] * 0.3 +
        X['total_of_special_requests'] * -0.4 +
        np.random.normal(0, 0.5, n_samples)
    )))
    
    y = (cancellation_prob > 0.5).astype(int)
    
    return X, y

def train_and_save_models():
    """Treina e salva modelos de demonstração"""
    # Criar dados
    X, y = create_demo_dataset()
    
    # Split dos dados
    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=0.3, random_state=42, stratify=y
    )
    
    # Normalizar
    scaler = StandardScaler()
    X_train_scaled = scaler.fit_transform(X_train)
    X_test_scaled = scaler.transform(X_test)
    
    # Treinar modelos
    models = {}
    results = {}
    
    # Regressão Logística
    lr = LogisticRegression(random_state=42, max_iter=1000)
    lr.fit(X_train_scaled, y_train)
    models['RL_Padrao'] = lr
    
    # KNN
    knn = KNeighborsClassifier(n_neighbors=5)
    knn.fit(X_train_scaled, y_train)
    models['KNN_Padrao'] = knn
    
    # SVM
    svm = SVC(probability=True, random_state=42)
    svm.fit(X_train_scaled, y_train)
    models['SVM_Padrao'] = svm
    
    # Avaliar modelos
    for name, model in models.items():
        y_pred = model.predict(X_test_scaled)
        y_proba = model.predict_proba(X_test_scaled)[:, 1]
        
        results[name] = {
            'Acurácia': accuracy_score(y_test, y_pred),
            'Precisão': precision_score(y_test, y_pred, zero_division=0),
            'Recall': recall_score(y_test, y_pred, zero_division=0),
            'F1-Score': f1_score(y_test, y_pred, zero_division=0),
            'AUC-ROC': roc_auc_score(y_test, y_proba),
            'Tempo Treino (s)': 0
        }
    
    # Salvar dados
    data_to_save = {
        'models': models,
        'X_train': X_train_scaled,
        'X_test': X_test_scaled,
        'y_train': y_train,
        'y_test': y_test,
        'results': results
    }
    
    joblib.dump(data_to_save, 'modelos_treinados.pkl')
    print("✅ Dados de demonstração e modelos salvos em 'modelos_treinados.pkl'")

if __name__ == "__main__":
    train_and_save_models()