for some function I need to check whether the passed series corresponds
to the variable defining the cross-sectional dimension of a panel
dataset. Here is some pseudo-code:
function void foo (series G)
if $datatype == 3 && G==is_crossdim_index_series(G)
--> Run pexpand()
--> Some other way to construct the return series
I was thinking of a function like is_crossdim_index_series() which
returns a boolean.
An alternative way I was thinking about is running:
<values($obsminor) == values(G)>
However, there is not guarantee that this works correctly as it may be
just luck that (conditional on the active sample) the distinct values of
G just coincide with the ones of $obsminor I guess.
Of course, the equivalent function for the time-dimension may be
Do you some other ideas?
Hi! I am wondering about the IRFs caculated from VEC models. In the IRF graphs, the values of the responses seem VERY small relative to a unit change in the impulse variable. Are they measured relative in units (relative to 1) or in standard deviations?
Good day por you:
Last days I'm making a graphs for a work, when I tried to graph a
bandplot with values tiny it appear wrong. I looked the code and find:
In the gnuplot code generated appear:
plot '-' using 1:($2-1.96*$3):($2+1.96*$3) notitle lc rgb "#efefef" w
0 notitle w lines lt 0, \
'-' using 1:2 title 'uAC' w lines lt 1
2000 1 0.2507141113
The value 1 after 2000 it's the central value for the band in the
generated code but my script code was:
gnuplot u$i --time-series --band=0, band$i, 1.96 --with-lines \
I check others graphs and always appear the 1 value, for big values
there isn't problem but for tiny values the band appear in the wrong place.
The coding gretl solution: change the value 1 for the first argument in
the --band option.
I'm not using gretlcli that often and so this is a purely decadent
question: For a while now we have had syntax coloring in the gretl
console; is it easy to enable the same thing for gretlcli on the shell?