##Larning library(nnet) library(tseries) #Logistic map x<-0 x[1]<-0.5 x[2]<-0.5 a<-3.8 for(i in 1:1000){ x[i+1]<-a*x[i]*(1-x[i]) } input<-cbind(x[1:(length(x)-1)]) output<-cbind(x[2:length(x)]) results<-nnet(input,output,size=5,lang=0.3,maxit=3000) plot(output[,1]) lag.plot(x,36) points(predict(results),col=2) j2<-predict(results) lag.plot(j2,36) ##Predicts a<-3.8 for(i in 1001:2000){ x[i+1]<-a*x[i]*(1-x[i]) } plot(x,col=2,type="l") l123<-0 l123b<-0 l123[1]<-x[1000] l123b[1]<-x[1000] a<-3.8 for(i in 1:1000){ l123[i+1]<-predict((results),l123[i]) l123b[i+1]<-a*l123b[i]*(1-l123b[i]) } #plot(l123,l123b) lag.plot(l123,lag=32)
線形合同法とラグプロット
x<-0 x[1]<-8 A<-12578 B<-32457 M<-1278 for(i in 1:10000){ x[i+1]<-(A*x[i]+B)%%M } lag.plot(x,lag=32) plot(x[1:100],type="l")
ランダムの中の複雑な相関の学習
x[1]<-1 x[2]<-1 x[3]<-1 for(i in 1:1000){ if(x[i-2]==0 && x[i-1] ==0 && x[i-3]==0){ x[i]<-1 }else{ x[i]<-rbinom(1,size=1,prob=0.5) } } plot(x,type="l") lag.plot(x,lag=36) x1<-x[1:(length(x)-3)] x2<-x[2:(length(x)-2)] x3<-x[3:(length(x)-1)] x4<-x[4:length(x)] resluts<-nnet(cbind(x1,x2,x3),cbind(x4),size=8,lang=0.3,maxit=3000) plot(x2) x3<-predict(resluts) points(x3,col=2) lag.plot(x3,lag=36) points(x3,col=2,type="l") lag.plot(x,lag=32) predict(resluts,c(0,0)) predict(resluts,c(0,1)) predict(resluts,c(1,0)) predict(resluts,c(1,1))