Monthly temperatures in Boulder, Colorado

Use this diagram to explain how to obtain forecasts for a seasonal time series,

Forecast  =   Seasonal effect   +   Trend forecast  +   Cyclical forecast

Use the checkboxes to add these three components separately to the bottom diagram. Explain how each can be extended into the future.

Seasonal effect
Forecast the same seasonal effects in the future.
Trend
Forecasts can be made easily from a linear or quadratic model.
Cyclical
Forecast using exponential smoothing or AR(1) model on the deseasonalised and detrended series

With all three checkboxes ticked, click on future months of the time series to see the resulting forecasts.

Since the trend and autocorrelation are negligible, forecasting temperatures from only the seasonal component would be equally accurate.


The data set gives the mean monthly temperature in Boulder Colorado between 1998 and 2007.