机器学习RF估算代码，加调参代码

使用RF估算浓度，价格，等，下面两个代码，第一个代码是寻找最优参数，第二个是预测。

话不多说直接上代码
RF_reserach.py

'''
@Project ：moxing 
@File    ：RF_reserach.py
@IDE     ：PyCharm 
@Author  ：11
@Date    ：2023/6/29 11:31 
'''
import numpy as np
import pandas as pd
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import GridSearchCV, train_test_split
import logging

# 配置日志输出
logging.basicConfig(filename='01month_rf_model_04csv.log', level=logging.INFO)
# 读取数据

data = pd.read_csv(r'G:\RF\normalized_month_12\01\04.csv')
# 定义特征和目标列
X = data.drop(columns=['CO2']) #目标列名
y = data['CO2']
# 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# 定义随机森林回归器
rf = RandomForestRegressor()


#定义参数网格
param_grid = {
     'n_estimators': [200, 300, 500,700,1000],  # 决策树数量
    'max_depth': [10, 20,],  # 最大树深度
     'min_samples_split': [2, 5,10,20],  # 节点分裂所需的最小样本数
   'min_samples_leaf': [1,2,4,8],  # 叶节点所需的最小样本数
  'max_features': ['sqrt']  # 最大特征数
 }

# 创建网格搜索对象
grid_search = GridSearchCV(estimator=rf, param_grid=param_grid, cv=5)

# 在训练集上进行参数搜索
grid_search.fit(X_train, y_train)
# 打印每个参数组合的结果
print("Grid Search Results:")
for params, mean_score, std_score in zip(grid_search.cv_results_['params'], grid_search.cv_results_['mean_test_score'], grid_search.cv_results_['std_test_score']):
    print(f"Parameters: {params}")
    print(f"Mean Score: {mean_score:.4f}")
    print(f"Std Score: {std_score:.4f}")
    print("--------------------")
# 打印最佳参数组合
print("Best Parameters:")
print(grid_search.best_params_)

# 打印最佳参数组合
best_params = grid_search.best_params_
logging.info("Best Parameters:")
logging.info(best_params)

# 使用最佳参数组合的模型进行预测
best_rf = grid_search.best_estimator_
y_pred = best_rf.predict(X_test)

# 在测试集上评估模型性能
mae = np.mean(abs(y_pred - y_test))
mse = np.mean((y_pred - y_test) ** 2)
rmse = np.sqrt(mse)
r2 = best_rf.score(X_test, y_test)

print("MAE:", mae)
print("MSE:", mse)
print("RMSE:", rmse)
print("R^2:", r2)
# 输出测试指标到日志
logging.info('测试集指标')
logging.info("MAE: %f", mae)
logging.info("MSE: %f", mse)
logging.info("RMSE: %f", rmse)
logging.info("R^2: %f", r2)

RF.py

'''
@Project ：moxing 
@File    ：RF.py
@IDE     ：PyCharm 
@Author  ：11
@Date    ：2023/6/29 13:41
'''
import numpy as np
import pandas as pd
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
import logging
# 配置日志输出
# logging.basicConfig(filename='02month_rf_model.log', level=logging.INFO)
# 读取数据
data = pd.read_csv(r'G:\RF\month_12\month_average\average11.csv')
# 定义特征和目标列
X = data.drop(columns=['co2'])
y = data['co2']

# 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# 定义随机森林回归器，并设置最优参数
rf = RandomForestRegressor(n_estimators=200, max_depth=30, min_samples_split=2,
                           min_samples_leaf=1, max_features='log2',bootstrap='False')

# 使用最优参数进行训练
rf.fit(X_train, y_train)

# 在测试集上进行预测
y_pred = rf.predict(X_test)

# 保存测试集的真值和预测值到 test_csv
test_data = pd.DataFrame({'y_true': y_test, 'y_pred': y_pred})
test_data.to_csv(r'G:\RF\month_12\month_average\RF\RF_test_csv_01.csv', index=False)

# 计算评估指标
mae = mean_absolute_error(y_test, y_pred)
mse = mean_squared_error(y_test, y_pred)
rmse = np.sqrt(mse)
# 计算相关系数
R = np.corrcoef(y_pred, y_test)[0, 1]
# 输出评估指标
print("MAE:", mae)
print("MSE:", mse)
print("RMSE:", rmse)
print("R:", R)

有问题的可以看作者名字q