Backcast : Backward Forecast


How can we use the smoothing functions to forecast data points before the start of the time series? I imaging if we reverse the time series (chronicle order), optimize the values of the smoothing function parameters, we can forecast for those dates. Am I missing something?

if not, can we have NumXL function does that for us? Like, if we specify the forecast horizon negative, it will infer the data point before the time series start?

Why, I would like to use current observations to imply/construct an outlook of the unknown past. For instance, given the last 10-15 years of temperature time series, how would be the temperature 20 years ago?


1 comment
  • Yes, I think this is possible, but bear in mind the time series model (e.g. ARMA, smoothing parameters values) won't be the same as the one used for forward forecast.

    Comment actions Permalink

Please sign in to leave a comment.

Didn't find what you were looking for?

New post