决策树回归2

# -*-coding:utf-8-*-
from matplotlib import pyplot as plt
from pandas.core.common import random_state
import time
from sklearn.datasets import load_diabetes
from sklearn.metrics import mean_squared_error, mean_absolute_error
from sklearn.model_selection import RandomizedSearchCV, train_test_split
from scipy.stats import randint
from sklearn.tree import DecisionTreeRegressor, plot_tree
import matplotlib.pyplot as plt
import numpy as np
import time
# 加载糖尿病数据集
diabetes_data = load_diabetes()
X = diabetes_data.data
y = diabetes_data.target

# 将数据集分割为训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# 定义参数网格
param_dist = {
    'criterion': ['mse', 'friedman_mse'],
    'max_depth': randint(5, 30),
    'min_samples_split': randint(2, 20),
    'min_samples_leaf': randint(5, 20),
    'max_features': ['auto', 'sqrt']
}

# 创建回归决策树模型
regressor = DecisionTreeRegressor(random_state=50)
start_time = time.time()
# 使用随机搜索进行参数调优
random_search = RandomizedSearchCV(estimator=regressor, param_distributions=param_dist, n_iter=100, cv=5, scoring='neg_mean_squared_error')
random_search.fit(X_train, y_train)

end_time = time.time()
# 输出最优参数和对应的评分
print("最优参数：", random_search.best_params_)
print("最优评分：", -random_search.best_score_)
training_time = end_time - start_time
print("训练时间：", training_time, "秒")
# 使用最优参数重新训练模型
best_regressor = random_search.best_estimator_
best_regressor.fit(X_train, y_train)
# 可视化决策树
plt.figure(figsize=(20,10))
plot_tree(best_regressor, filled=True, feature_names=diabetes_data.feature_names)
plt.draw()
plt.show()
# 预测测试集结果
y_pred = best_regressor.predict(X_test)

# 计算均方误差（MSE）
mse = mean_squared_error(y_test, y_pred)
print("均方误差（MSE）：", mse)

# 计算均方根误差（RMSE）
rmse = mean_squared_error(y_test, y_pred, squared=False)
print("均方根误差（RMSE）：", rmse)

# 计算平均绝对误差（MAE）
mae = mean_absolute_error(y_test, y_pred)
print("平均绝对误差（MAE）：", mae)

# 计算决定系数（R^2）
r_squared = best_regressor.score(X_test, y_test)
print("决定系数（R^2）：", r_squared)


# 绘制预测值和实际值的散点图
plt.rcParams['font.sans-serif'] = ['SimSun', 'Songti SC', 'Microsoft YaHei']
plt.scatter(y_test, y_pred)
plt.xlabel("实际值")
plt.ylabel("预测值")
# 计算并绘制趋势线或回归线
trendline_x = np.linspace(np.min(y_test), np.max(y_test), 100)
trendline_y = trendline_x
plt.plot(trendline_x, trendline_y, color='red')
plt.show()