模型筛选-回归模型

#!/usr/bin/env python

-- coding:utf-8 --

import os
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.metrics import mean_squared_error
from sklearn.metrics import r2_score
from sklearn.metrics import mean_absolute_error
from sklearn.metrics import mean_absolute_percentage_error
plt.rcParams[‘font.sans-serif’] = [‘SimHei’] #显示中文
plt.rcParams[‘axes.unicode_minus’]=False

###########1.数据准备##########
filename=r’/data/data202401_202404.csv’
path = os.path.dirname(os.path.realpath(file))
csvpath= os.path.dirname(os.path.realpath(file)) + filename
data = pd.read_csv(csvpath, parse_dates=True, index_col=‘datetime’)
data[‘实际功率’]=data[‘realpower’].map(lambda x: x if x> 0 else 0)
data=data[‘实际功率’]
train,test=data[:‘2024/4/1 0:15’],data[‘2024/4/1 0:15’:]

def process(dataset):
#观看过去时间窗口 过去多少天
past_history_size = 16
#预测未来值n天
future_target = 1
x = []
y = []
dataset=dataset.values
for i in range(len(dataset)-past_history_size-future_target+1):
x.append(dataset[i:i+past_history_size])
y.append(dataset[i+past_history_size:i+past_history_size+future_target])
x,y= np.array(x),np.array(y)
x = x.reshape([-1,past_history_size])
print(‘*’*100)
print(x.shape,y.shape)
return x ,y

trainX,trainY = process(train)
testX,testY = process(test)
print(testX.shape,testY.shape)

###########2.模型训练通法##########
def try_different_method(model):
# 创建对象并填入将样本改成 最高degree次幂
model.fit(trainX,trainY)
result = model.predict(testX)
mse=mean_squared_error(testY,result)
rmse = np.sqrt(mean_squared_error(testY,result))
r2score=r2_score(testY,result)
mae=mean_absolute_error(testY,result)
mape= mean_absolute_percentage_error(testY,result)
print(‘mse’,mse)
print(‘rmse’,rmse)
print(‘mae’,mae)
print(‘mape’,mape)
print(‘r2score’,r2score)

plt.rcParams['font.sans-serif']=['SimHei'] #用来正常显示中文标签
plt.plot(testY.reshape(-1,1),label="真实数据")
plt.plot(result.reshape(-1,1),label="预测值")
plt.legend(loc=1)
plt.title(str(model)[:-2])
plt.savefig(path+'/pictures/'+str(model)[:-2]+'.png')
plt.close()

return mse,rmse,mae,mape,r2score

###########3.具体方法选择##########
####3.1决策树回归####
from sklearn import tree
model_DecisionTreeRegressor = tree.DecisionTreeRegressor()
####3.2线性回归####
from sklearn import linear_model
model_LinearRegression = linear_model.LinearRegression()
####3.3SVM回归####
from sklearn import svm
model_SVR = svm.SVR()
####3.4KNN回归####
from sklearn import neighbors
model_KNeighborsRegressor = neighbors.KNeighborsRegressor(n_neighbors=16)
####3.5随机森林回归####
from sklearn import ensemble
model_RandomForestRegressor = ensemble.RandomForestRegressor(n_estimators=16)#这里使用16个决策树
####3.6Adaboost回归####
from sklearn import ensemble
model_AdaBoostRegressor = ensemble.AdaBoostRegressor(n_estimators=50)#这里使用50个决策树
####3.7GBRT回归####
from sklearn import ensemble
model_GradientBoostingRegressor = ensemble.GradientBoostingRegressor(n_estimators=100)#这里使用100个决策树
####3.8Bagging回归####
from sklearn.ensemble import BaggingRegressor
model_BaggingRegressor = BaggingRegressor()
####3.9ExtraTree极端随机树回归####
from sklearn.tree import ExtraTreeRegressor
model_ExtraTreeRegressor = ExtraTreeRegressor()
####3.10ARD贝叶斯ARD回归
model_ARDRegression = linear_model.ARDRegression()
####3.11BayesianRidge贝叶斯岭回归
model_BayesianRidge = linear_model.BayesianRidge()
####3.12TheilSen泰尔森估算
model_TheilSenRegressor = linear_model.TheilSenRegressor()
####3.13RANSAC随机抽样一致性算法
model_RANSACRegressor = linear_model.RANSACRegressor()
####3.14Lightgbm梯度提升算法
from lightgbm import LGBMRegressor
model_LGBMRegressor= LGBMRegressor()
####3.15CatBoost梯度提升算法
from catboost import CatBoostRegressor
model_CatBoostRegressor=CatBoostRegressor()

###########4.具体方法调用部分##########
#决策树回归结果
try_different_method(model_DecisionTreeRegressor)
#线性回归结果
#try_different_method(model_LinearRegression)
#SVM回归结果
#try_different_method(model_SVR)
#KNN回归结果
#try_different_method(model_KNeighborsRegressor)
#随机森林回归结果
#try_different_method(model_RandomForestRegressor)
#Adaboost回归结果
#try_different_method(model_AdaBoostRegressor)
#GBRT回归结果
#try_different_method(model_GradientBoostingRegressor)
#Bagging回归结果
#try_different_method(model_BaggingRegressor)
#极端随机树回归结果
#try_different_method(model_ExtraTreeRegressor)
#贝叶斯ARD回归结果
#try_different_method(model_ARDRegression)
#贝叶斯岭回归结果
#try_different_method(model_BayesianRidge)
#泰尔森估算回归结果
#try_different_method(model_TheilSenRegressor)
#随机抽样一致性算法
#try_different_method(model_RANSACRegressor)
#Lightgbm梯度提升算法
#try_different_method(model_LGBMRegressor)
#CatBoost梯度提升算法
#try_different_method(model_CatBoostRegressor)

###########5.所有方法调用部分##########
method_list=[model_DecisionTreeRegressor,model_LinearRegression,model_SVR,model_KNeighborsRegressor,
model_RandomForestRegressor,model_AdaBoostRegressor,model_GradientBoostingRegressor,
model_BaggingRegressor,model_ExtraTreeRegressor,model_ARDRegression,model_BayesianRidge,
model_TheilSenRegressor,model_RANSACRegressor,model_LGBMRegressor,model_CatBoostRegressor]
result_dict={}
def train_all(method_list,path):
for method in method_list:
mse,rmse,mae,mape,r2score=try_different_method(method)
result_dict[type(method).name] = [mse, rmse, mae, mape, r2score]

df = pd.DataFrame.from_dict(result_dict, orient='index', columns=['mse', 'rmse', 'mae', 'mape', 'r2score'])
# 将'method'作为行索引
df.index.name = 'method'
# 保存DataFrame到CSV
df.to_csv(f'{path}/result/screen_model_result.csv', encoding='utf-8')

train_all(method_list,path)
#print(result_dict)