Series 0 not found using grc1leg2

[Florian Schneider]

Dear all, I am using Stata 14 on Windows 10. The following code is supposed to create a combined graph of several scatter plots for two variables. I used almost the same code several times for other graphs and it worked. However, this time I always get the error message "Series 0 not found" after all single graphs for the respective countries have been generated, i.e. no combined graph is created (I only get an empty combined graph). After spending several hours trying to fix that problem, I have no more idea...

levelsof country, local(levels)

local graph ""

foreach i of local levels {
    summarize net_tradevalue_m if country == "`i'", meanonly
    local y_min = `r(min)'
    local y_max = `r(max)'
    
    local y_min_r = round(`y_min')
    local y_max_r = round(`y_max')
    
    scatter net_tradevalue_m instal if country == "`i'", ///
    xtitle("Number of items installed") ///
    ytitle("Import value (US$) net of re-exports") ///
    yscale(range(`y_min' `y_max')) ylabel(`y_min_r' `y_max_r') ///
    title("`i'") name(g_`i', replace)
    local graph "`graph' g_`i'"
}

grc1leg2 `graph', xtob1title ytol1title ytsize(vsmall) labsize(vsmall)

* Example generated by -dataex-.
year country net_tradevalue_m instal
2005 "AE"   .596557    3
2007 "AE"    .57997    2
2008 "AE"  1.334496    2
2012 "AE"    .82767    4
2013 "AE"   2.23105    2
2014 "AE"  3.857373    0
2015 "AE"  2.996899    0
2016 "AE"  3.456476    0
2017 "AE" 10.282882    0
2018 "AE"  6.563038    0
2019 "AE"  3.967778    0
1996 "AR"  2.930427    0
1997 "AR"  2.566672    0
1998 "AR"  2.713423    0
1999 "AR"  1.757973   20
2000 "AR"   .820514   50
2001 "AR"  1.731663   40
2002 "AR"   .433283   29
2003 "AR"   .693137   33
2004 "AR"   .537529   17
2005 "AR"  1.334015   65
2006 "AR"  1.607395   36
2007 "AR"  5.205558  141
2008 "AR" 13.348274  150
2009 "AR"  2.953936   45
2010 "AR"  8.285274   96
2011 "AR" 22.273457  407
2012 "AR" 15.632359  180
2013 "AR" 13.275552  181
2014 "AR" 10.507074    0
2015 "AR" 14.350166    0
2016 "AR" 11.726218    0
2017 "AR" 16.001951    0
2018 "AR"  9.526949    0
2019 "AR" 15.542093    0
1996 "AT"  9.284512  277
1997 "AT" 11.324429  232
1998 "AT"  11.18832  145
1999 "AT" 16.971384  273
2000 "AT"  37.53419  320
2001 "AT" 16.574034  330
2002 "AT" 16.691013  670
2003 "AT"  14.51144  365
2004 "AT" 24.916763  545
2005 "AT"  25.72232  485
2006 "AT" 35.112324  498
2007 "AT"  38.49554  621
2008 "AT"  46.18709  638
2009 "AT"  41.32523  508
2010 "AT"  36.71757  496
2011 "AT"  43.99167  628
2012 "AT"  40.30826  835
2013 "AT"   40.7037  720
2014 "AT"  51.64431  898
2015 "AT"  48.09217  987
2016 "AT"   73.0638 1686
2017 "AT"     81.19 1641
2018 "AT"  82.17607 1504
2019 "AT"  74.40486 1475
1996 "AU"  4.763878  250
1997 "AU" 10.905708  526
1998 "AU"  6.894873  347
1999 "AU"  8.601764  180
2000 "AU"  8.956798  440
2001 "AU" 10.291573  270
2002 "AU"  13.03586  421
2003 "AU" 17.182116  569
2004 "AU" 28.043364  652
2005 "AU"  29.79912  890
2006 "AU" 19.709654  719
2007 "AU"  26.95258  734
2008 "AU"  20.39395  781
2009 "AU"  8.937055  399
2010 "AU" 13.016978  624
2011 "AU"  17.47799  690
2012 "AU"   22.5622 1214
2013 "AU"  13.40527  323
2014 "AU" 11.591695    0
2015 "AU" 13.899048    0
2016 "AU"  16.18698    0
2017 "AU" 19.643173    0
2018 "AU"  28.74417    0
2019 "AU" 18.894983    0
2003 "BA"     .0334    0
2004 "BA"   .166554    0
2005 "BA"   .074738    0
2006 "BA"   .260394    0
2007 "BA"   .535262    0
2008 "BA"   .677682    0
2009 "BA"  1.532421    0
2010 "BA"   .381561    2
2011 "BA"  3.805619    2
2012 "BA"   .817377    0
2013 "BA"   .587407    0
2014 "BA"  1.019831    0
2015 "BA"   .604327    0
2016 "BA"   .752781    0
2017 "BA"  1.305355    0
2018 "BA"  1.188319    0
2019 "BA"  2.205264    0
end

It would be great, if someone could help me. Thanks a lot!

[Nick Cox]

What's wrong with starting from

 
 scatter net_tradevalue_m instal, by(country) xtitle("Number of items installed") ytitle("Import value (US$) net of re-exports")

?
grc1leg2 is community-contributed, so you are asked to explain where it comes from (FAQ Advice #12).

[Florian Schneider]

The problem is that I need individual scales for the y-axis and x-axis for every country. Your code leads to a uniform scale for every graph such that the output graph is dominated by the country with the highest values, while for all other countries no pattern is discernable.

I heard of the command grc1leg2 in this topic on statalist: https://www.statalist.org/forums/for...one-legend-box

I simpliy installed grc1leg2 by typing in: search grc1leg2 and then clicking on install (package grc1leg2 from http://digital.cgdev.org/doc/stata/MO/Misc).

[Nick Cox]

The words "starting from" were carefully chosen. The positive point that follows is that you can ask for y and/or x scales to vary through suboptions of by().

The dataex example is #1 is very helpful, but you've edited out code that is deliberately part of the output to allow the example to run correctly. I've reinstated that.

Also alphabetical order is sometimes helpful for look-up, but often not. I've included a demonstration of myaxis to put the countries in one of many possible better orders. The latter is from SSC and discussed
at https://www.statalist.org/forums/for...e-or-graph-use

Very likely you have many more countries, but the problems of that will bite regardless of how you combine the graphs.

* Example generated by -dataex-.
clear 
input year str2 country net_tradevalue_m instal
2005 "AE"   .596557    3
2007 "AE"    .57997    2
2008 "AE"  1.334496    2
2012 "AE"    .82767    4
2013 "AE"   2.23105    2
2014 "AE"  3.857373    0
2015 "AE"  2.996899    0
2016 "AE"  3.456476    0
2017 "AE" 10.282882    0
2018 "AE"  6.563038    0
2019 "AE"  3.967778    0
1996 "AR"  2.930427    0
1997 "AR"  2.566672    0
1998 "AR"  2.713423    0
1999 "AR"  1.757973   20
2000 "AR"   .820514   50
2001 "AR"  1.731663   40
2002 "AR"   .433283   29
2003 "AR"   .693137   33
2004 "AR"   .537529   17
2005 "AR"  1.334015   65
2006 "AR"  1.607395   36
2007 "AR"  5.205558  141
2008 "AR" 13.348274  150
2009 "AR"  2.953936   45
2010 "AR"  8.285274   96
2011 "AR" 22.273457  407
2012 "AR" 15.632359  180
2013 "AR" 13.275552  181
2014 "AR" 10.507074    0
2015 "AR" 14.350166    0
2016 "AR" 11.726218    0
2017 "AR" 16.001951    0
2018 "AR"  9.526949    0
2019 "AR" 15.542093    0
1996 "AT"  9.284512  277
1997 "AT" 11.324429  232
1998 "AT"  11.18832  145
1999 "AT" 16.971384  273
2000 "AT"  37.53419  320
2001 "AT" 16.574034  330
2002 "AT" 16.691013  670
2003 "AT"  14.51144  365
2004 "AT" 24.916763  545
2005 "AT"  25.72232  485
2006 "AT" 35.112324  498
2007 "AT"  38.49554  621
2008 "AT"  46.18709  638
2009 "AT"  41.32523  508
2010 "AT"  36.71757  496
2011 "AT"  43.99167  628
2012 "AT"  40.30826  835
2013 "AT"   40.7037  720
2014 "AT"  51.64431  898
2015 "AT"  48.09217  987
2016 "AT"   73.0638 1686
2017 "AT"     81.19 1641
2018 "AT"  82.17607 1504
2019 "AT"  74.40486 1475
1996 "AU"  4.763878  250
1997 "AU" 10.905708  526
1998 "AU"  6.894873  347
1999 "AU"  8.601764  180
2000 "AU"  8.956798  440
2001 "AU" 10.291573  270
2002 "AU"  13.03586  421
2003 "AU" 17.182116  569
2004 "AU" 28.043364  652
2005 "AU"  29.79912  890
2006 "AU" 19.709654  719
2007 "AU"  26.95258  734
2008 "AU"  20.39395  781
2009 "AU"  8.937055  399
2010 "AU" 13.016978  624
2011 "AU"  17.47799  690
2012 "AU"   22.5622 1214
2013 "AU"  13.40527  323
2014 "AU" 11.591695    0
2015 "AU" 13.899048    0
2016 "AU"  16.18698    0
2017 "AU" 19.643173    0
2018 "AU"  28.74417    0
2019 "AU" 18.894983    0
2003 "BA"     .0334    0
2004 "BA"   .166554    0
2005 "BA"   .074738    0
2006 "BA"   .260394    0
2007 "BA"   .535262    0
2008 "BA"   .677682    0
2009 "BA"  1.532421    0
2010 "BA"   .381561    2
2011 "BA"  3.805619    2
2012 "BA"   .817377    0
2013 "BA"   .587407    0
2014 "BA"  1.019831    0
2015 "BA"   .604327    0
2016 "BA"   .752781    0
2017 "BA"  1.305355    0
2018 "BA"  1.188319    0
2019 "BA"  2.205264    0
end

set scheme s1color 

scatter net_tradevalue_m instal , by(country, yrescale)

myaxis better=country, sort(median net_tradevalue_m) descending 

scatter net_tradevalue_m instal , by(better, yrescale)

[Andrew Musau]

For your specific problem, it appears that it is triggered by including the option -labsize()-. You will want to alert Mead Over, the author of grc1leg2, to take a look.

* Example generated by -dataex-.
clear 
input year str2 country net_tradevalue_m instal
2005 "AE"   .596557    3
2007 "AE"    .57997    2
2008 "AE"  1.334496    2
2012 "AE"    .82767    4
2013 "AE"   2.23105    2
2014 "AE"  3.857373    0
2015 "AE"  2.996899    0
2016 "AE"  3.456476    0
2017 "AE" 10.282882    0
2018 "AE"  6.563038    0
2019 "AE"  3.967778    0
1996 "AR"  2.930427    0
1997 "AR"  2.566672    0
1998 "AR"  2.713423    0
1999 "AR"  1.757973   20
2000 "AR"   .820514   50
2001 "AR"  1.731663   40
2002 "AR"   .433283   29
2003 "AR"   .693137   33
2004 "AR"   .537529   17
2005 "AR"  1.334015   65
2006 "AR"  1.607395   36
2007 "AR"  5.205558  141
2008 "AR" 13.348274  150
2009 "AR"  2.953936   45
2010 "AR"  8.285274   96
2011 "AR" 22.273457  407
2012 "AR" 15.632359  180
2013 "AR" 13.275552  181
2014 "AR" 10.507074    0
2015 "AR" 14.350166    0
2016 "AR" 11.726218    0
2017 "AR" 16.001951    0
2018 "AR"  9.526949    0
2019 "AR" 15.542093    0
1996 "AT"  9.284512  277
1997 "AT" 11.324429  232
1998 "AT"  11.18832  145
1999 "AT" 16.971384  273
2000 "AT"  37.53419  320
2001 "AT" 16.574034  330
2002 "AT" 16.691013  670
2003 "AT"  14.51144  365
2004 "AT" 24.916763  545
2005 "AT"  25.72232  485
2006 "AT" 35.112324  498
2007 "AT"  38.49554  621
2008 "AT"  46.18709  638
2009 "AT"  41.32523  508
2010 "AT"  36.71757  496
2011 "AT"  43.99167  628
2012 "AT"  40.30826  835
2013 "AT"   40.7037  720
2014 "AT"  51.64431  898
2015 "AT"  48.09217  987
2016 "AT"   73.0638 1686
2017 "AT"     81.19 1641
2018 "AT"  82.17607 1504
2019 "AT"  74.40486 1475
1996 "AU"  4.763878  250
1997 "AU" 10.905708  526
1998 "AU"  6.894873  347
1999 "AU"  8.601764  180
2000 "AU"  8.956798  440
2001 "AU" 10.291573  270
2002 "AU"  13.03586  421
2003 "AU" 17.182116  569
2004 "AU" 28.043364  652
2005 "AU"  29.79912  890
2006 "AU" 19.709654  719
2007 "AU"  26.95258  734
2008 "AU"  20.39395  781
2009 "AU"  8.937055  399
2010 "AU" 13.016978  624
2011 "AU"  17.47799  690
2012 "AU"   22.5622 1214
2013 "AU"  13.40527  323
2014 "AU" 11.591695    0
2015 "AU" 13.899048    0
2016 "AU"  16.18698    0
2017 "AU" 19.643173    0
2018 "AU"  28.74417    0
2019 "AU" 18.894983    0
2003 "BA"     .0334    0
2004 "BA"   .166554    0
2005 "BA"   .074738    0
2006 "BA"   .260394    0
2007 "BA"   .535262    0
2008 "BA"   .677682    0
2009 "BA"  1.532421    0
2010 "BA"   .381561    2
2011 "BA"  3.805619    2
2012 "BA"   .817377    0
2013 "BA"   .587407    0
2014 "BA"  1.019831    0
2015 "BA"   .604327    0
2016 "BA"   .752781    0
2017 "BA"  1.305355    0
2018 "BA"  1.188319    0
2019 "BA"  2.205264    0
end

levelsof country, local(levels)

local graph ""

foreach i of local levels {
    summarize net_tradevalue_m if country == "`i'", meanonly
    local y_min = `r(min)'
    local y_max = `r(max)'
    
    local y_min_r = round(`y_min')
    local y_max_r = round(`y_max')
    
    scatter net_tradevalue_m instal if country == "`i'", ///
    xtitle("Number of items installed") ///
    ytitle("Import value (US$) net of re-exports") ///
    yscale(range(`y_min' `y_max')) ylabel(`y_min_r' `y_max_r') ///
    title("`i'") name(g_`i', replace)
    local graph "`graph' g_`i'"
}

grc1leg2 `graph', xtob1title ytol1title ytsize(vsmall)

[Florian Schneider]

Dear Andrew, thanks for your response! So it worked out for you when leaving out labsize()? Mysterious... I tried it without option labsize and even without any option, but the problem does persist and I get the error message "Series 0 not found".

[Florian Schneider]

Nick Cox: Many thanks for your hint! I was able to build a nice graph.

However, I am still puzzled why my code seems to be non-functional only in this specific context.

[Andrew Musau]

Dear Andrew, thanks for your response! So it worked out for you when leaving out labsize()? Mysterious... I tried it without option labsize and even without any option, but the problem does persist and I get the error message "Series 0 not found".

I had an older version. Updating to the latest, I can replicate your error with the code in #5. So I would say that this occurs within the last 2 updates.

. which grc1leg2
c:\ado\plus\g\grc1leg2.ado
*! version 1.42 29Mar2021, by Mead Over
*! based on -grc1leg- version 1.0.5  02jun2010, by Vince Wiggins

[Mead Over]

Thanks to Florian Schneider and Andrew Musau for flagging this problem. I think Nick Cox 's suggestion to Florian to use gr ..., by(..., yrescale xrescale) is a good solution. And his new program -myaxis- is cool, Meanwhile I am trying to figure out this problem with -grc1leg2-. As I emphasize in the help file for -grc1leg2-, its approach to creating a combined graph with a single legend is best used only when -gr ..., by()- won't work.

[Mead Over]

Andrew Musau is correct that the bug that Florian Schneider found dates back to version 1.41 which I distributed this year on March 26. In the process of fixing this bug, I've discovered that -grc1leg2- has difficulty editing certain characteristics of the keys of a legend that is too simple, i.e. that has too few keys. I don't expect this problem to arise for most users because (a) most will not attempt to edit the characteristics of legend keys and (b) a legend with only a single key is typically uninformative and should usually be suppressed. In the new version 1.5 of -grc1leg2-, I've added a rubric to the help file documenting this "Known Issue" and suggesting work-arounds. As always, please contact me if you see remaining issues with the program - or if you can suggest how to trap the serset error "series # not found".

[Daniel Gross]

I don't know if this is helpful to Mead Over or others, and it's perhaps not surprising, but I get this error with the original -grc1leg- as well. Only it shows up as "series 2 not found" (rather than series 0, above), and seems to be an issue when there are few keys provided by the user, as previously noted. After this error is tripped, then seems to cause errors in any subsequent uses of -graph combine- or commands like -grc1leg- that wrap it. I cleared and re-ran my code after changing -grc1leg- to -graph combine- and it worked fine, but that sacrifices the -grc1leg- functionality.