时间序列组合预测

install.packages("zoo") install.packages("forecast") install.packages("tseries") install.packages("e1071") library("e1071") library("tseries") library("zoo") library("forecast") dt = read.csv("C://Users//yelang//Desktop//matlabyewu//股票预测//数据.csv") #读入数据 airline2 <- dt[1:55,2] airts <- ts(airline2,start=c(2011,2),frequency=12) plot.ts(airts) airdiff <- diff(diff(airts, differences=1)) plot.ts(airdiff) adf.test(airdiff, alternative="stationary", k=0) #检验序列平稳性 acf(airdiff, lag.max=20) acf(airdiff, lag.max=20,plot=FALSE) pacf(airdiff, lag.max=20) pacf(airdiff, lag.max=20,plot=FALSE) auto.arima(airdiff,trace=T) airarima1 <- arima(airline2,order=c(3,1,0),seasonal=list(order=c(3,1,0),period=12),method="ML") airarima1 airarima2 <- arima(airline2,order=c(3,1,0),seasonal=list(order=c(1,1,0),period=12),method="ML") airarima2 airarima3 <- arima(airline2,order=c(1,1,0),seasonal=list(order=c(1,1,0),period=12),method="ML") airarima3 airforecast <- forecast.Arima(airarima3,h=16,level=c(99.5)) airforecast plot.forecast(airforecast) #画预测图 ycforecast <- forecast.Arima(airarima1,h=16) #预测后面16个 write.csv(tq[1],file="C://Users//yelang//Desktop//matlabyewu//股票预测//yc1.csv")#写出数据 sub1=dt[56:68,2]-tq[1] sub1=as.matrix(sub1) #转换成矩阵 #4个数据循环 xindt=matrix(NA,4,length(sub1)-6) for (i in 1: 10){ xindt[,i]=c(sub1[i],sub1[i+1],sub1[i+2],sub1[i+3]) } inputData<-cbind(t(xindt[,1:9]),sub1[5:13]) set.seed(100) # for reproducing results rowIndices <- 1 : nrow(inputData) # prepare row indices sampleSize <- 0.8 * length(rowIndices) # training sample size trainingRows <- sample (rowIndices, sampleSize) # random sampling trainingData <- inputData[trainingRows, ] # training data testData <- inputData[-trainingRows, ] # test data #tuned <- tune.svm(V5 ~., data = trainingData, type="eps-regression" , gamma = 10^-1, cost = 10^(1:2)) # tune #summary (tuned) # to select best gamma and cost svmfit <- svm (V5 ~ ., data = trainingData, type="eps-regression" , kernel = "radial", cost = 1000, gamma=0.0001, scale = FALSE) # radial svm, scaling turned OFF print(svmfit) predict(svmfit, trainingData[,1:4]) compareTable <- cbind(testData[,5], predict(svmfit, testData[,1:4])) # comparison table View(compareTable) yc1=predict(svmfit, t(xindt)) write.csv(yc1,file="C://Users//yelang//Desktop//matlabyewu//股票预测//yc2.csv")#写出数据 #预测后面第1个 h1=sub1[10:13,1] h1=as.matrix(h1) row.names(h1)<-c("V1","V2","V3","V4") th1=predict(svmfit, t(h1)) #预测后面第二个 h2=c(sub1[11:13,1],th1) h2=as.matrix(h2) row.names(h2)<-c("V1","V2","V3","V4") th2=predict(svmfit, t(h2)) #预测后面第三个数 h3=c(sub1[12:13,1],th1,th2) h3=as.matrix(h3) row.names(h3)<-c("V1","V2","V3","V4") th3=predict(svmfit, t(h3)) # http://www.cnblogs.com/xuancaoyy/p/5535909.html c参考网址