[Gretl-users] Re: EXTRACTION OF FORECAST DATA FOR USE IN EXCEL
On Tue, 26 May 2026, Brian Revell wrote:
Hi
when estimating either with OLS, or Univariate ARmax model
specifications (Gretl 2026b)there is a problem extracting the
Further Analysis Forecast data for download into Excel for improved
graphi presentation. Copying the forecast output produces output
which appears as below. Whilst visually pleasing, unfortunately the
columns are not separable for the forecasts 2025-30. No matter how
one tries to parse the 2025-2030 block of data, it remains
unseparable via the Excel command process of DATA Text to columns.
I take your point. In the window that shows the forecasts there's a
"Save as" button that offers the options plain text, LaTeX (if
relevant) and RTF. It would be handy if a CSV option were added. We'll
see about doing that (in which case we could also offer copying as
CSV).
Any thoughts on workarounds for the meantime, to avoid having to
reinput the forecasts manually into the Excel spreadsheet.
The toolbar at the top of the forecast window has a "+" ("Add to
dataset") button. You can use that to add the forecasts and standard
errors as series. Then from the main window you can view the series --
but note, if you shortened the sample range so as to produce
out-of-sample forecasts you'll have to re-establish the full range to
see everything. And in the viewer window for series you have the
option to copy as CSV.
This workaround will not actually give you the 95% interval, but once
the data are in a spreadsheet it's easy enough to add and subtract 2
standard errors. If 2 isn't a good enough approximation, make note of
the t-statistic shown at the top of the forecast printout.
Allin Cottrell