1、数据划分
R语言实现:
将专家样本划分为测试样本和训练样本,随机选取20%为测试样本,剩下来的作为训练样本。
data = read.csv(file="model.csv")
colnames(data) = c("time","userid","ele_ind","loss_ind","alarm_ind","class") #数据命名
head(data)
输出结果:
set.seed(1234) #设置随机种子
ind = sample(2,nrow(data),replace = TRUE,prob = c(0.8,0.2)) #定义序列ind,随机抽取1和2,1的个数占80%,2的个数占20%
traindata = data[ind==1,] #训练样本
testdata = data[ind==2,] #测试样本
write.csv(traindata,"traindata.csv",row.names = FALSE)
write.csv(testdata,"testdata.csv",row.names = FALSE) # 保存到当期工作目录下
Python实现:
import pandas as pd
from random import shuffle #导入随机函数shuffle,用来打乱数据
data = pd.read_excel("model.xls")
data = data.as_matrix() #将表格转换为矩阵
shuffle(data)
p = 0.8
train = data[:int(len(data)*p),:] #训练集
test = data[int(len(data)*p):,:] #测试集
2、建模-神经网络
R语言实现:
traindata =transform(traindata,class = as.factor(class)) #将class列转换成因子类型
library(nnet)
nnet.model = nnet(class~ele_ind+loss_ind+alarm_ind,traindata,size=10,decay=0.05) #设定神经网络的输入节点为ele_ind,loss_ind,alarm_ind,输出节点为class,数据为traindata,隐层节点数为10,权值的衰减参数为0.05
confusion = table(traindata$class,predict(nnet.mo