Hi John, a couple of years ago, I've written a package "plot_scatter_label". Unfortunately, it's not on the package server any more -- I don't know why. But here is the code plus an example: <hansl> clear set verbose off function string AddLabel (series Y, series X, strings LABEL) matrix XY = {Y}~{X} string S = "" loop i=1..rows(XY) -q s = sprintf("set label $i '%s' at %.4f-0.1, %.4f-0.2 centre font ',10' ; ",\ LABEL[$i], XY[i,2],XY[i,1]) S += s endloop return S end function open mrw.gdt --quiet smpl 1 3 strings C = array(3) C[1] = "Algeria" C[2] = "Angola" C[3] = "Benin" string GP = AddLabel(gdpgrow,school, C) # Actual Plot list lplot = gdpgrow school plot lplot literal set term png font 'Helvetica,12' lw 1 literal set nokey literal set title '' literal set xrange[0:] literal set linetype 1 lc rgb 'blue' pt 7 ps 1 literal @GP end plot --output=display </hansl> Am 01.06.2022 07:55 schrieb John Paravantis: > I am talking about adding the values of a qualitative variable, e.g. > country codes, as labels. > > As an example from my megacity research, imagine a scatter plot of > carbon footprint (y-axis) vs city population (x-axis). In such a plot, > it would be very useful if the data points were labeled with the > megacity name. > > So I am suggesting that two separate improvements be considered for > gretl: > > (1) Add the capability of handling qualitative (i.e. text) variables, > something that is sorely missing right now. I have to manually delete > columns containing text from Excel files that I want to open in gretl, > which is inconvenient. If anything else, gretl could ignore such > columns, not give an error and refuse to open such a file. > (2) Add the capability of adding a qualitative variable as labels of > the points of a scatter plot. > > Many thanks to all for your thoughts, > John > > On Wed, Jun 1, 2022 at 1:00 AM <gretl-users-request(a)gretlml.univpm.it&gt; > wrote: > >> Am 31.05.2022 um 00:19 schrieb John Paravantis: >>> If I may get back to my suggestion, I understand that it may be >> possible >>> to add data labels via commands, but it would appear to me that >> it would >>> not be too difficult to add an option of selecting the values of >> a >>> variable as labels, since the option of adding labels to >> individual >>> points is already built in in the edit dialog of a scatter plot. >> >> John, let me ask the more fundamental question, why would one want >> the >> values as labels? I mean in a scatter plot the values are on the >> axes >> and the location of the data point communicates those values >> already. >> OK, sometimes one wants to see the exact digits and then a kind of >> mouse-over tooltip effect is nice. But really printing/displaying >> the >> values right now does not sound to me like a natural GUI feature. >> >> cheers >> sven > > _______________________________________________ > Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it > To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it > Website: > https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/

Am 01.06.22 um 13:30 schrieb Sven Schreiber: Ah, thanks for the reminder! Yes, good idea. Artur

On Wed, 1 Jun 2022, John Paravantis wrote: Sven has given you good answers. I'll just add a couple of points. * If you have an Excel file with string-valued variables that gretl won't load, please post it (offlist, perhaps) so we can see what's going wrong. Note that in gretl a single series cannot mix string and numeric values (but a series containing nothing but city names should be fine). * Gretl's plotting apparatus supports showing observation markers as labels, but suppose the strings of interest do not take the form of observation markers -- they're just a string-valued series. In that case you can add obs markers with the same content. In the example below I assume that the series named "city" is string valued: <hansl> strings S = array($nobs) loop i=1..$nobs S[i] = city[i] endloop markers --from-array=S </hansl> Allin