BP神经网络预测

setwd("/users/xuwei/desktop/R/chapter5/示例程序")
myData<-read.csv("sales_data.csv",header=F)[,2:5]
head(myData)
library(nnet)
model1<-nnet(V5~.,data=myData,size=6,decay=5e-4,maxit=1000)
pred<-predict(model1,myData[,1:3],type="class")
(p=sum(as.numeric(pred==myData$V5))/nrow(myData))
[1] 0.7647059
table(myData$V5,pred)
      pred
       high low
  high   14   4
  low     4  12
prop.table(table(myData$V5,pred),1)
      pred
            high       low
  high 0.7777778 0.2222222
  low  0.2500000 0.7500000

检测样本34个，预测正确的14＋12＝26个，预测准确率26／34