R code：one step ahead forecast

1. Download from Quandl the quarterly S&P 500 Index ranging from 1988.Q1 to 2013.Q4.
2. Divide the series into the in-sample estimation window (observations up to and includ-
ing 2009.Q4) and the out-of-sample forecasting window (observations beginning from
2010.Q1 and onwards).
3. Produce the point and the interval forecasts using the simple exponential smoothing,
the Holt's linear trend, and the Holt-Winters additive seasonal models. Plot the three
graphs.
4. Estimate the rst-order autoregression, i.e. AR(1) process, and calculate the uncondi-
tional mean and the unconditional variance measures.
5. Produce the point and the interval forecasts from the estimated autoregression. Plot
the graph.
6. Within the rolling forecasting environment, generate one-step-ahead forecasts through-
out the out-of-sample set using the considered four models.
7. Calculate aand report the out-of-sample mean absolute error (MAE) and root mean
squared error (RMSE) measures. Which model is preferred?

初学R code 第六步与第七步没有思路，请帮帮忙。谢谢啦！

我想问下楼主，这个第六点用中文怎么翻译，主要是one-step-ahead forecasts， out-of-sample，这两个我不太会。Within the rolling forecasting environment, generate one-step-ahead forecasts through-out the out-of-sample set using the considered four models.