基于随机森林算法的空气质量预测

源数据集

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import confusion_matrix
from sklearn.metrics import accuracy_score
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import mean_squared_error
from sklearn.tree import DecisionTreeRegressor

# 读取csv文件
data = pd.read_csv('DataSet.csv')

# 转化数据格式 因为co只有1.x 精确到一位会使精度丢失
data['CO'] = data.iloc[:,2:3].round(1)
# data = data.iloc[:,2:].round(0)
data['AQI'] = data['AQI'].round(0)
data.iloc[:,3:] = data.iloc[:,3:].round(0)

x_train,x_validation,y_train,y_validation = train_test_split(data.iloc[:,2:].round(0),data.iloc[:,1].round(0),test_size=0.2)
# 拆的是训练集和验证集 测试集不参与
# test_size 指验证集所占数据的百分比

# 先拆分在进行标准化 
# 先标准化在进行拆分 x 因为标准化会带着验证集的数据一起 不准确
from sklearn.preprocessing import StandardScaler
# 标准化 生成规则
stdScale = StandardScaler().fit(x_train)
# 将规则用于训练集
x_train = stdScale.transform(x_train)
# 将规则用于验证集:（验证集-训练集均值）/ 训练集方法
x_validation = stdScale.transform(x_validation)

# 建立模型CART
dtc = DecisionTreeClassifier(criterion = 'gini',
                             random_state = 0,
#                              若节点要拆，最少样本要有6个
                             min_samples_split = 6,
#                              叶子节点包含的样本数
                             min_samples_leaf = 3,
                             max_features = None)
# 训练模型
dtc.fit(x_train,y_train)
# 测试结果
result_y = dtc.predict(x_validation)

# 建立模型
dtc = DecisionTreeClassifier()
# 训练模型
dtc.fit(x_train,y_train)
# 测试结果
result_y = dtc.predict(x_validation)

y_validation.index.tolist()

accuracy_score(result_y,y_validation)
confusion_matrix(result_y,y_validation)

rfc = RandomForestClassifier(n_estimators = 100,
                             random_state=0,
                             min_samples_split=6,
                             min_samples_leaf=2,
                             max_features=None,
                             oob_score = True)

# 训练模型
rfc.fit(x_train,y_train)
# 测试结果
result_y1 = rfc.predict(x_validation)

accuracy_score(result_y1,y_validation)
rfc.oob_score_

# 随即森林回归模型
regressor = DecisionTreeRegressor(random_state=0,
                                  min_samples_split=6,
                                  min_samples_leaf=2,
                                  max_features=None)
regressor.fit(x_train,y_train)
result_y2 = regressor.predict(x_validation).round(2)

result_y2 = pd.Series(result_y2.tolist())

print(mean_squared_error(result_y2,y_validation))
from sklearn.metrics import r2_score
# 随即森林分类R21
R21 = r2_score(result_y1,y_validation)
# 随即森林回归R22
R22= r2_score(result_y2,y_validation)
R22

# 随即森林回归算法 画图
plt.scatter(result_y2.index.tolist(),y_validation,color = 'g',label='实际值',s=10)
plt.scatter(result_y2.index.tolist(),result_y2,color='y',label='预测值',s=10)
plt.xlabel('样本点')
plt.ylabel('AQI')
plt.legend();
plt.show()