Problem creating a combined bar and line graph with two y-axes

[Aiman Mitra]

Hi,

I am attempting to create a combined bar and line graph with two y-axes including a legend and colour format. I have struggled putting together code from my university sources for these graphs individually.

The dataset I am working with is as follows:

input int date double(sv tv svtv)
2021 263265.01367681153 1884107.5491802376 .13972929188217326
2020 160562.117150004 1247802.0562035448 .12867595172788582
2019 176471.19955512544 1749111.8072808338 .10089189199944128
2018 170093.77284257952 1804252.3167105296 .09427382814876496
2017 52161.34900366189 1402845.4179146306 .03718253510868019
2016 114215.2414812902 1684493.7934715918 .06780389570085785
2015 80283.8120504003 1874024.0737563872 .042840331228763476
2014 58058.12532963137 1554142.3946453538 .037357017947431966
2013 40682.652479267206 1019954.9128099392 .03988671652866298
2012 38677.2364842603 914746.105465388 .04228193621505805
2011 57203.51218648255 967624.8487086741 .059117448526484664
2010 35043.662845142186 837891.7568109324 .041823615712034835
2009 29106.356725849208 656567.4335384262 .04433110026336043
2008 19632.641050937582 889188.2822891638 .022079284491237878
2007 108563.48645410615 1625505.757096285 .0667875127357519
2006 45838.282629804686 1511020.1616175605 .030335983459502218
2005 37117.18074119836 1183223.5713459835 .03136954134456219
2004 31706.67899895087 889931.028397809 .035628243074111166
2003 29608.788667924702 579999.2549263801 .051049701213293754
2002 11840.867066330742 482330.77073760197 .02454926740050848
2001 13351.644165197888 792923.65140603 .01683849906800289
2000 55384.067105144255 1553569.4440576113 .03564955999681108

I am trying to graph sv with a bar graph and svtv with a line graph, would anyone know how I can attempt this? Thanks in advance!

[Nick Cox]

Such combinations are quite popular but in my experience hard to get right. You don't show the code you tried but some trickery includes

* putting the line on top of the bar, not vice versa

* matching colours of axis title and labels to what is shown in the plot region (harder if you are restricted on colours you can use).

I also suggest trying multiline from SSC, as explained at https://www.statalist.org/forums/for...ailable-on-ssc

clear 
input int date double(sv tv svtv)
2021 263265.01367681153 1884107.5491802376 .13972929188217326
2020 160562.117150004 1247802.0562035448 .12867595172788582
2019 176471.19955512544 1749111.8072808338 .10089189199944128
2018 170093.77284257952 1804252.3167105296 .09427382814876496
2017 52161.34900366189 1402845.4179146306 .03718253510868019
2016 114215.2414812902 1684493.7934715918 .06780389570085785
2015 80283.8120504003 1874024.0737563872 .042840331228763476
2014 58058.12532963137 1554142.3946453538 .037357017947431966
2013 40682.652479267206 1019954.9128099392 .03988671652866298
2012 38677.2364842603 914746.105465388 .04228193621505805
2011 57203.51218648255 967624.8487086741 .059117448526484664
2010 35043.662845142186 837891.7568109324 .041823615712034835
2009 29106.356725849208 656567.4335384262 .04433110026336043
2008 19632.641050937582 889188.2822891638 .022079284491237878
2007 108563.48645410615 1625505.757096285 .0667875127357519
2006 45838.282629804686 1511020.1616175605 .030335983459502218
2005 37117.18074119836 1183223.5713459835 .03136954134456219
2004 31706.67899895087 889931.028397809 .035628243074111166
2003 29608.788667924702 579999.2549263801 .051049701213293754
2002 11840.867066330742 482330.77073760197 .02454926740050848
2001 13351.644165197888 792923.65140603 .01683849906800289
2000 55384.067105144255 1553569.4440576113 .03564955999681108
end 

set scheme s1color 
gen sv2 = sv/1000
gen tv2 = tv/1000 
label var sv2 "sv (000)"
label var tv2 "tv (000)"

local opts1 yaxis(1) yla(, ang(h) labcolor(black) axis(1)) fcolor(black*0.2) lcolor(black) ytitle(, color(black) axis(1))
local opts2 yaxis(2) yla(, ang(h) labcolor(red) axis(2)) lw(medthick) lcolor(red) ytitle(, color(red) axis(2)) 


twoway bar sv2 date, `opts1' || line tv2 date, `opts2' legend(off) xtitle("") name(G1, replace)

multiline sv2 tv2 date , name(G2, replace) xtitle("") xla(, grid glw(thin)glc(gs8))

[Nick Cox]

You can go further with colouring axis line and tick marks.

[Aiman Mitra]

Hi Nick! Thank you so much for the help!

[Nick Cox]

It seems possible that plotting both on log scale would help and that recalling that log(sv / tv) = log sv - log tv would be instructive.