arima：如何拟合ARIMA时间序列？

Question

m7 = arima(lill,order=c(0,0,1),
          seasonal=list(order=c(1,0,0),period=22),
          xreg=data.frame(lpGDP))

preds = predict(m7,n.ahead = 1, newxreg = 1)

There are 329 observations in lill object. How can I predict the last observation 328, instead of 330 observation? Thank you.

Answer 1

You don't need to call predict for prediction of observed data. You can do:

fitted_values <- lill - m7$residuals

This is the fitted ARIMA model. To inspect the 328th value, do

fitted_values[328]

I don't have your data, so I use R's built-in data set LakeHuron as a toy demonstration.

fit <- arima(LakeHuron, order = c(2,0,0), xreg = time(LakeHuron) - 1920)
fitted_values <- LakeHuron - fit$residuals
ts.plot(LakeHuron)  ## observed time series (black)
lines(fitted_values, col = 2)  ## fitted time series (red)