pso优化.rarpso优化.rarpso优化.rar

import numpy as np from sklearn.model_selection import train_test_split from sklearn.ensemble import GradientBoostingRegressor, AdaBoostRegressor from xgboost import XGBRegressor from lightgbm import LGBMRegressor from sklearn.metrics import mean_squared_error from pyswarm import pso import pandas as pd import math from data_read import date_process train_x,train_y=date_process() train_x=train_y.reshape(train_x.shape[0],-1) X_train, X_test, y_train, y_test = train_test_split(np.array(train_x), np.array(train_y), test_size=0.2,shuffle=False,random_state=15) # 定义粒子群优化的目标函数 def objective_function(params): # 解析超参数 xgb_learning_rate, xgb_n_estimators, xgb_max_depth = params[:3] ab_n_estimators, ab_learning_rate = params[3:5] lgb_learning_rate, lgb_n_estimators, lgb_max_depth = params[5:] # 初始化回归模型 xgb_model = XGBRegressor(learning_rate=xgb_learning_rate, n_estimators=int(xgb_n_estimators), max_depth=int(xgb_max_depth)) ab_model = AdaBoostRegressor(n_estimators=int(ab_n_estimators), learning_rate=ab_learning_rate) lgb_model = LGBMRegressor(learning_rate=lgb_learning_rate, n_estimators=int(lgb_n_estimators), max_depth=int(lgb_max_depth)) # 训练模型 xgb_model.fit(X_train, y_train) ab_model.fit(X_train, y_train) lgb_model.fit(X_train, y_train) # 预测 y_pred_xgb = xgb_model.predict(X_test) y_pred_ab = ab_model.predict(X_test) y_pred_lgb = lgb_model.predict(X_test) # 合并预测结果 y_pred_ensemble = (y_pred_xgb + y_pred_ab + y_pred_lgb) / 3.0 # 计算均方根误差作为目标函数值 mse = mean_squared_error(y_test, y_pred_ensemble) return mse # 定义超参数搜索范围 lower_bound = [0.01, 50, 3, 50, 0.01, 0.01, 50, 3] upper_bound = [0.02, 51, 4, 51, 0.02, 0.02, 51, 4] # upper_bound = [0.1, 200, 10, 200, 0.1, 0.1, 200, 10] # 使用粒子群优化算法进行超参数优化 best_params, _ = pso(objective_function, lower_bound, upper_bound, swarmsize=10, maxiter=50) # 输出最佳超参数 print("Best Hyperparameters:", best_params) # 使用最佳超参数训练模型 best_learning_rate_xgb, best_n_estimators_xgb, best_max_depth_xgb = best_params[:3] best_n_estimators_ab, best_learning_rate_ab = best_params[3:5] best_learning_rate_lgb, best_n_estimators_lgb, best_max_depth_lgb = best_params[5:] best_xgb_model = XGBRegressor(learning_rate=best_learning_rate_xgb, n_estimators=math.ceil(best_n_estimators_xgb), max_depth=math.ceil(best_max_depth_xgb)) best_ab_model = AdaBoostRegressor(n_estimators=math.ceil(best_n_estimators_ab), learning_rate=best_learning_rate_ab) best_lgb_model =LGBMRegressor(learning_rate=best_learning_rate_lgb, n_estimators=math.ceil(best_n_estimators_lgb), max_depth=math.ceil(best_max_depth_lgb)) best_xgb_model.fit(X_train, y_train) best_ab_model.fit(X_train, y_train) best_lgb_model.fit(X_train, y_train) # 进行预测 y_pred_xgb = best_xgb_model.predict(X_test) y_pred_ab = best_ab_model.predict(X_test) y_pred_lgb = best_lgb_model.predict(X_test) # 合并预测结果 y_pred_ensemble = (y_pred_xgb + y_pred_ab + y_pred_lgb) / 3.0 from metra import metric mae, mse, rmse, mape, mspe,r2=metric(np.array(y_pred_ensemble), np.array(y_test)) print('mae, mse, rmse, mape, mspe,r2') print(mae, mse, rmse, mape, mspe,r2) # 设置Seaborn样式 import seaborn as sns import matplotlib.pyplot as plt plt.figure(figsize=(10, 6)) plt.plot(y_test, label='True Values', color='b') plt.plot(y_pred_ensemble, label='Predicted Values', color='r', linestyle='--') plt.xlabel('Sample Index') plt.ylabel('Value') plt.title('True Values vs Predicted Values') plt.legend() plt.grid(True) plt.show()