On Tue, 15 May 2012, Sven Schreiber wrote:
not important, but here goes after having postponed it for ages:
1. Graphing the cross-correlogram seems to rely on the ordering of the
variables in the dataset. Could that be made configurable?
It's configurable if you (a) don't select two series in the
main window, but (b) invoke the cross-correlogram via the menu
2. the "semilog" fitted line option for time-series plots:
first why is
it called *semi*, if it's a fitted trend to the log.
Well, the trend case is specific to time-series; in general
it's a graph of the fit log(y) = a + b*x (not b*log(x)).
Also, trying it on LRM in denmark.gdt there is no difference
between linear and this semilog. (How) Does gretl know that
LRM is already in logs?
No, the plot shows the level of the variable in question plus
the exponentiated log-fit. It turns out that in this case the
linear and exponentiated-log fits are practically