# Importando bibliotecas necessárias
# Código utilizando 3 modelos: DecisionTreeRegressor, RandomForestRegressor e GradientBoostingRegressor
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor, GradientBoostingRegressor
from sklearn.datasets import fetch_california_housing
from sklearn.metrics import mean_squared_error, r2_score
# Carregar dataset
california = fetch_california_housing()
df = pd. DataFrame(california.data, columns=california.feature_names)
df['PRICE'] = california.target
# Dividindo os dados em treino e teste
X = df.drop('PRICE', axis=1)
y = df['PRICE']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)
# Definir hiperparâmetros para otimização
param_grid = {
'max_depth': [3, 5, 7, 10],
'min_samples_split': [2, 5, 10],
'min_samples_leaf': [1, 2, 4]
}
# ------Treino do modelo DecisionTreeRegressor-------- #
# Dividir os dados em treino e teste (70% treino, 30% teste)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)
# Treinar o LLM no modelo Decision Tree
tree_model = DecisionTreeRegressor() # 'LLM' here seems to be a typo, it should be 'model'
tree_model.fit(X_train, y_train)
# Aplicar GridSearch para encontrar os melhores hiperparâmetros
grid_search = GridSearchCV(DecisionTreeRegressor(), param_grid, cv=5, scoring='neg_mean_squared_error')
grid_search.fit(X_train, y_train)
# Avaliando o modelo Decision Tree
y_pred_tree = tree_model.predict(X_test)
mse_tree = mean_squared_error(y_test, y_pred_tree)
r2_tree = r2_score(y_test, y_pred_tree)
print(f"Decision Tree - MSE: {mse_tree:.2f}, R²: {r2_tree:.2f}")
# -------Criando e treinando modelos ensemble-------- #
rf_model = RandomForestRegressor(n_estimators=200, max_depth=10, random_state=42)
gb_model = GradientBoostingRegressor(n_estimators=200, learning_rate=0.05, max_depth=5, random_state=42)
rf_model.fit(X_train, y_train)
gb_model.fit(X_train, y_train)
# Fazendo previsões
y_pred_rf = rf_model.predict(X_test)
y_pred_gb = gb_model.predict(X_test)
# Avaliando os modelos
mse_rf = mean_squared_error(y_test, y_pred_rf)
r2_rf = r2_score(y_test, y_pred_rf)
mse_gb = mean_squared_error(y_test, y_pred_gb)
r2_gb = r2_score(y_test, y_pred_gb)
# Exibindo métricas
print(f"Random Forest - MSE: {mse_rf:.2f}, R²: {r2_rf:.2f}")
print(f"Gradient Boosting - MSE: {mse_gb:.2f}, R²: {r2_gb:.2f}")
# Visualização dos resultados
plt.figure(figsize=(12, 5))
plt.subplot(1, 2, 1)
sns.scatterplot(x=y_test, y=y_pred_rf, alpha=0.6)
plt.plot([min(y_test), max(y_test)], [min(y_test), max(y_test)], '--r', linewidth=2)
plt.xlabel("Valores Reais")
plt.ylabel("Valores Preditos")
plt.title("Random Forest")
plt.subplot(1, 2, 2)
sns.scatterplot(x=y_test, y=y_pred_gb, alpha=0.6, color='red')
plt.plot([min(y_test), max(y_test)], [min(y_test), max(y_test)], '--r', linewidth=2)
plt.xlabel("Valores Reais")
plt.ylabel("Valores Preditos")
plt.title("Gradient Boosting")
plt.show()
