-xtline- with smoothing

[Anton Ivanov]

Hello!

I am exploring new visualizations with Stata, such as a line graph with arrows following an example presented by @Asjad Naqvi on Medium. Say, my data looks like this:

* Example generated by -dataex-. For more info, type help dataex
clear
input float date int country2 float(count_total_week cases_pop pos_test cumulative_test) str13 country
3206   0   155   2  .012903226  .012903226 "Voluntary"    
3207   0  2432  16  .006578947  .019482173 "Voluntary"    
3208   0  6039  12  .001987084   .02146926 "Voluntary"    
3209   0  7789  28  .003594813   .02506407 "Voluntary"    
3210   0 17312  58  .003350277   .02841435 "Voluntary"    
3211   0 18119  47  .002593962   .03100831 "Voluntary"    
3212   0 23537  81   .00344139    .0344497 "Voluntary"    
3213   0 27981 112  .004002716   .03845242 "Voluntary"    
3214   0 31134  98  .003147684    .0416001 "Voluntary"    
3215   0 29917 110  .003676839   .04527694 "Voluntary"    
3216   0 31244 159  .005088977   .05036592 "Voluntary"    
3217   0 31942 174  .005447373   .05581329 "Voluntary"    
3206   1   404   2  .004950495  .004950495 "Unvax all"    
3207   1   721  12   .01664355  .021594046 "Unvax all"    
3208   1  3129  15  .004793864   .02638791 "Unvax all"    
3209   1  4510  28  .006208426  .032596335 "Unvax all"    
3210   1  5723  25 .0043683383  .036964674 "Unvax all"    
3211   1  6604  24  .003634161   .04059884 "Unvax all"    
3212   1 11085  35   .00315742   .04375625 "Unvax all"    
3213   1 12930  40  .003093581   .04684984 "Unvax all"    
3214   1 13876  37 .0026664746   .04951631 "Unvax all"    
3215   1 15762  78   .00494861   .05446492 "Unvax all"    
3216   1 17063  91  .005333177    .0597981 "Unvax all"    
3217   1 16185 111  .006858202    .0666563 "Unvax all"    
3206   2   391   7  .017902814  .017902814 "Unvax stu/fac"
3207   2   360   4   .01111111  .029013924 "Unvax stu/fac"
3208   2   854   4   .00468384  .033697765 "Unvax stu/fac"
3209   2  1954   3  .001535312  .035233077 "Unvax stu/fac"
3210   2  3426  13 .0037945125   .03902759 "Unvax stu/fac"
3211   2  3916  18  .004596527   .04362412 "Unvax stu/fac"
3212   2  3803  11  .002892453   .04651657 "Unvax stu/fac"
3213   2  5925  23 .0038818566   .05039843 "Unvax stu/fac"
3214   2  6328  30  .004740834   .05513926 "Unvax stu/fac"
3215   2  6399  54  .008438818   .06357808 "Unvax stu/fac"
3216   2  8154 127  .015575178   .07915325 "Unvax stu/fac"
3217   2  8049 105    .0130451   .09219836 "Unvax stu/fac"
3206   3  1715  11  .006413994  .006413994 "Everyone"     
3207   3  3416  17  .004976581  .011390575 "Everyone"     
3208   3  3004  10  .003328895   .01471947 "Everyone"     
3209   3  1731   2 .0011554015  .015874872 "Everyone"     
3210   3  2115  10  .004728132  .020603003 "Everyone"     
3211   3  2394  10 .0041771093  .024780113 "Everyone"     
3212   3  3785   7 .0018494056   .02662952 "Everyone"     
3213   3  4311  15  .003479471   .03010899 "Everyone"     
3214   3  5707  27  .004731032   .03484002 "Everyone"     
3215   3  6068  25 .0041199736      .03896 "Everyone"     
3216   3  5853  26 .0044421665   .04340216 "Everyone"     
3217   3  6243  39  .006246997   .04964916 "Everyone"     
3208   4    28   0           0           0 "Selective"    
3209   4   496   0           0           0 "Selective"    
3210   4  1653   5 .0030248035 .0030248035 "Selective"    
3211   4  1367   3 .0021945867   .00521939 "Selective"    
3212   4  2816   6 .0021306819  .007350072 "Selective"    
3213   4  3731  23  .006164567   .01351464 "Selective"    
3214   4  3260  13   .00398773   .01750237 "Selective"    
3215   4  3939  15  .003808073   .02131044 "Selective"    
3216   4  5221  51  .009768244  .031078685 "Selective"    
3217   4  5310  43  .008097928   .03917661 "Selective"    
3206 999 30560 161  .005268325  .005268325 "No report"    
3207 999 34032 212  .006229431  .011497756 "No report"    
3208 999 56555 289   .00511007  .016607825 "No report"    
3209 999 51274 231 .0045052073   .02111303 "No report"    
3210 999 54119 236 .0043607606   .02547379 "No report"    
3211 999 51783 179  .003456733  .028930526 "No report"    
3212 999 64097 178  .002777041  .031707566 "No report"    
3213 999 82817 301 .0036345194  .035342086 "No report"    
3214 999 81826 344 .0042040427   .03954613 "No report"    
3215 999 84020 470  .005593906   .04514004 "No report"    
3216 999 80720 511  .006330526   .05147056 "No report"    
3217 999 85794 629  .007331515   .05880208 "No report"    
end

I then attempt to run the following:

sort country date
gen cases_pop_smooth =. 
levelsof country2, local(lvls)

foreach x of local lvls {
    lowess  cases_pop date if country2==`x', bwid(0.15) gen(temp`x') nograph
    replace cases_pop_smooth =temp`x'  if country2==`x'
    drop temp`x'
}
xtset country2 date
xtline cases_pop_smooth, overlay legend(off)

But for some reason, The loop makes no changes in the cases_pop_smooth variable. What am I doing wrong?

[William Lisowski]

Your bandwidth is apparently too small, given the number of observations (at least in your example data) and apparently lowess does not tell the user when this is the case. Try starting with bwid(0.8) which is the default and reduce the value from there.

[Anton Ivanov]

William Lisowski That makes sense, thank you!