http://tjo.hatenablog.com/entry/2013/07/12/184704
を参考に。
データ: sentdex.com/GBPUSD.zip
tb <- read.table("GBPUSD1d.txt",sep=',',colClasses=c("character","numeric","numeric")) df <- data.frame(strptime(tb$V1,"%Y%m%d%M%S"),tb$V2,tb$V3) names(df) <- c("time","bid","ask") >head(df) time bid ask 1 2013-05-01 1.55358 1.55371 2 2013-05-01 1.55357 1.55369 3 2013-05-01 1.55353 1.55367 4 2013-05-01 1.55349 1.55367 5 2013-05-01 1.55349 1.55367 6 2013-05-01 1.55348 1.55367 bids <- df$bids bids_model<-auto.arima(bids,max.p=2,max.q=2,stepwise=T,trace=T) plot(forecast(bids_model,level=c(50,95),h=10000))
上がるか下がるか、何も言ってないように思える。
追記:ここの方法が参考になりそう
http://www.rdatamining.com/examples/time-series-forecasting