Multiple line plot of long data

[Igor Paploski]

Hello all,

I’m working with a dataset in the long format. My observations are grouped within ids (id) and weeks (week), the combination of id and week make an observation unique. I have a probabilities (prob) that I would like to plot as a line graph for different observations. This would be done stratified (separate graphs) for different groups (group).

Please find below a mock dataset similar to what my real data looks like:

clear
input long id str10 week float prob byte group
1 "1/10/2020"   .14303502 2
1 "1/3/2020"    .12902983 2
1 "12/20/2019"  .04932104 2
1 "12/27/2019"  .04299045 2
2 "1/10/2020"    .1206926 1
2 "1/3/2020"    .05439886 1
2 "12/20/2019" .032277483 1
2 "12/27/2019"  .05520606 1
3 "1/10/2020"    .1378371 1
3 "1/3/2020"     .1603722 1
3 "12/20/2019"  .06625363 1
3 "12/27/2019"   .1458157 1
4 "1/10/2020"   .11175546 1
4 "1/3/2020"     .0779216 1
4 "12/20/2019"   .1137393 1
4 "12/27/2019" .064635806 1
5 "1/10/2020"   .09516551 1
5 "1/3/2020"     .1502394 1
5 "12/20/2019"  .12006584 1
5 "12/27/2019"  .12520134 1
end

I’ve tried using linkplot – from SSC (sencode also came from SSC):

gen date = date(week, "MDY")
sort date
sencode week, gen(week2)
 
linkplot prob week2 if group==1, link(id)

But I wasn’t able to display different lines (ids) using different colors, nor to display a legend that says which line is which id (if lines had different colors, then this legend would display that the navy line is id 1, the maroon is id 2 and so on). Other than the color and legend, this is the kind of graph I have in mind - it displays the prob variation through time for each id - this is the essence of the graph I want.

Am I overlooking something simple here? Does anyone have suggestions on how to proceed on this? I appreciate any tips or hints.

Cheers

[Scott Merryman]

You could try -xtline-:

drop if group == 2
xtset id week2
xtline prob, overlay recast(connected)

[Igor Paploski]

As expected, there was something simple. Thank you very much Scott!