colour coding in two-way dropline plot

[Mamun Huda]

Hello,
I am trying to distinguish droplines in a two-way dropline plot using different colour codes based on a categorical variable. Please note that I have different number of droplines in each category. Is it possible to do it in Stata?

Thanks in advance for your response and support.

[Nick Cox]

Yes, that is likely to be possible. separate is likely to be the key. To get precise advice you'll need to give your exact command and ideally a data example, neither of which is yet evident.

[Mamun Huda]

Many thanks, Nick for your kind response.

I will explore the "separate" key. However, I am also sending you my example dataset:

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte(region event) int month double index
1 1 101 -.04456543
1 1 102 -.22366977
1 1 103 -.18224127
1 1 104 -.18520155
1 1 105 -.25422327
1 1 106 -.03274019
1 1 107   -.214101
1 1 108 -.15150441
1 1 109 -.05893304
1 2 110 -.17668299
1 2 111  .03202023
1 2 112 -.07304322
1 2 113 -.18920047
1 2 114  -.0293457
1 2 115 -.07255485
1 2 116 -.10823589
1 2 117  .02739052
1 2 118 -.16033876
1 3 119 -.14558274
1 3 120 -.26880398
1 3 121 -.10420506
1 3 122  .04819746
1 3 123  -.1582941
1 3 124 -.07386343
1 3 125 -.20657758
1 3 126 -.16347356
1 3 127 -.14267572
1 4 128 -.20825162
1 4 129  -.1708047
1 4 130 -.11546738
1 4 131  -.1311501
1 4 132 -.13549459
1 4 133 -.22564544
1 4 134   .0210928
1 4 135 -.00756734
1 4 136 -.02268052
1 4 137  -.1559068
1 4 138 -.13051796
1 4 139 -.13820574
1 4 140 -.13526144
1 4 141 -.05863849
1 4 142  .00654885
1 4 143 -.18984016
1 4 144  -.1525872
1 4 145 -.22168383
1 4 146 -.12905128
1 4 147 -.09187518
1 1 148 -.04403891
2 4 101 -.02632621
2 4 102 -.07098256
2 4 103 -.35597657
2 4 104 -.22720844
2 4 105 -.19867589
2 4 106 -.13407439
2 4 107 -.34761308
2 4 108 -.29313611
2 1 109 -.29486544
2 1 110 -.18954702
2 1 111 -.08406134
2 1 112 -.17567455
2 1 113 -.26873224
2 1 114 -.05506746
2 1 115 -.10025514
2 1 116 -.29882877
2 1 117 -.24819362
2 1 118  -.2640195
2 1 119 -.23581635
2 1 120 -.36479868
2 1 121 -.35368169
2 1 122  .06226837
2 2 123 -.18890692
2 2 124 -.30127199
2 2 125 -.23595627
2 2 126  -.2606628
2 2 127 -.23865505
2 2 128 -.26654745
2 2 129 -.42581543
2 2 130  -.2620678
2 2 131 -.25116891
2 2 132 -.18500453
2 2 133 -.23495972
2 2 134  .07698295
2 2 135 -.14358356
2 2 136 -.13876035
2 2 137 -.15299878
2 2 138 -.23099192
2 2 139 -.23454866
2 3 140 -.36762659
2 3 141 -.17897339
2 3 142 -.09265568
2 3 143  -.1954482
2 3 144   -.310174
2 3 145 -.25596743
2 3 146  -.2285826
2 3 147 -.21706971
2 3 148 -.14608445
end

my code was:

twoway dropline index month, hori xline(0) by(region)

to generate a horizontal dropline plot with two panels (by region). Here I would like to highlight (colour code) different dropline group by the variable "event" in my dataset.

Thanks again.

[Nick Cox]

Thanks for your code and data example. I did this:

set scheme s1color 

separate index, by(event) veryshortlabel 
twoway  dropline index1 month, yline(0) ///
     || dropline index2 month,          ///
     || dropline index3 month, by(region, col(1) note("")) /// 
ytitle(unexplained index) xtitle("") legend(row(1)) yla(0 "0" 0.2 -0.2 -0.4, format("%02.1f") ang(h)) /// 
xtick(95.5(12)155.5, tlength(*2)) xla(101.5 "1968" 113.5 "1969" 125.5 "1970" 137.5 "1971" 149.5 "1972", tlength(*0.2) tlc(none))

On your question see also https://www.stata-journal.com/sjpdf....iclenum=gr0023 on separate, particularly the otherwise undocumented veryshortlabel option.

I dropped the horizontal option, which seems odd for this kind of data. It's your graph and you can put it back.

Some other technique shown, for which see also https://www.stata-journal.com/sjpdf....iclenum=gr0030

[Nick Cox]

#4 I left out index4 from the graph code. That was just a mistake.

[Mamun Huda]

Hi Nick,

Thank you so much for your help, and sharing the code and links. It helps me to solve my problem.

Thank you once again!