Doubt with daily dates and x-axis

[Juan Carlos]

Greetings,

I am working with a database and I have two doubts:

1) is there any way that in the x-axis I can only show the year (leaving not visible the month and days). Some time Nick helped me to modify the axis, but I have tried that way and the axis does not look good at all. I used tlabel(, format(%dM/CY)), but as I don't know the STATA number I can't manipulate the axis with the dates.

I would like to do something like what I show you in the attached image

2) the base that I imported already had the dates and STATA recognized it well, but I would like to clone the variable where I have the date and from there pass it to the pure format of STATA and with it I can work in the part of the axes (for example 1/2/2012 -> 18994).

Thanks for your attention colleagues!!!

clear
input int fecha double tdc
18994 13.9342
18995 13.6882
18996 13.7144
18997 13.7409
18998 13.7228
19001 13.7437
19002 13.6204
19003 13.6592
19004 13.5982
19005 13.6263
19008 13.5401
19009 13.4621
19010 13.3267
19011 13.2516
19012  13.224
19015 13.1383
19016 13.1712
19017 13.1319
19018 12.9325
19019 12.9504
19022 13.0063
19023 13.0077
19024   12.89
19025 12.8038
19026  12.712
19030 12.6472
19031 12.6833
19032   12.72
19033 12.7857
19036 12.7392
19037 12.7753
19038 12.7658
19039 12.8822
19040 12.7856
19043 12.6793
19044 12.7408
19045 12.8296
19046 12.8294
19047 12.8675
19050 12.8779
19051 12.8575
19052 12.7891
19053 12.7764
19054 12.7723
19057 12.8339
19058 12.9777
19059 12.9469
19060  12.784
19061 12.6299
19064  12.715
19065 12.6336
19066 12.6719
19067 12.6591
19068 12.6645
19072 12.6935
19073 12.6833
19074  12.833
19075 12.8035
19078 12.6795
19079 12.6706
19080 12.8039
19081 12.8489
19082 12.8093
19085 12.7691
19086 12.7317
19087 12.8155
19092 12.9902
19093 13.1568
19094 13.1324
19095 13.0736
19096 13.1458
19099 13.2168
19100 13.0816
19101 13.1322
19102 13.2288
19103 13.1171
19106 13.2087
19107 13.1392
19108 13.1667
19109 13.2093
19110 13.0141
19113 12.9942
19115 12.9629
19116 12.9725
19117 13.1344
19120 13.1659
19121 13.3374
19122  13.515
19123  13.447
19124 13.4929
19127 13.6739
19128 13.7302
19129 13.7747
19130 13.7994
19131 13.7809
19134 13.7935
19135 13.7203
19136 14.0335
19137  14.004
19138 14.0156
end
format %tdnn/dd/CCYY fecha

tsset fecha

tsline tdc

[Nick Cox]

Juan is alluding to

https://www.statalist.org/forums/for...is-double-time

where the problem was axis labels for quarterly data.

Note also his thread

https://www.statalist.org/forums/for...g-ticks-x-axis

where it's the same problem for monthly data.

The point is simple but was aired systematically in https://www.stata-journal.com/articl...rticle=gr0030: with a moderately (or extremely) long time series, axis labels for every date would usually be a (very) bad idea, and indeed for say quarterly or monthly data, the preferred labels would usually start (and possibly stop) with showing years.

The question in #1 is about daily dates and the sample graph in #1 shows year labels for 2012/2023. So one approach, consistent with the 2007 paper just cited. is to put ticks at each 1 January (31 December if you really prefer) and labels without ticks at each 1 July (or any date close to that).

mylabels and myticks from the Stata Journal can help here. Here is a silly example to make the point, code first and then a copy of results.

clear 
set obs 11 
gen date = mdy(7, 1, 2011 + _n)
gen foo = 42 
format date %td 

gen year = year(date)
su year, meanonly 

mylabels 2012/2023, myscale(mdy(7, 1, @)) local(labels)
myticks 2012/2023, myscale(mdy(1,1,@)) local(ticks)

line foo date, xla(`labels', noticks) xtic(`ticks', tlength(*3)) xtitle("")

. clear 

. set obs 11 
Number of observations (_N) was 0, now 11.

. gen date = mdy(7, 1, 2011 + _n)

. gen foo = 42 

. format date %td 

. 
. gen year = year(date)

. su year, meanonly 

. 
. mylabels 2012/2023, myscale(mdy(7, 1, @)) local(labels)
19175 "2012" 19540 "2013" 19905 "2014" 20270 "2015" 20636 "2016" 21001 "2017" 21366 "2018" 21731
"2019" 22097 "2020" 22462 "2021" 22827 "2022" 23192 "2023"

. myticks 2012/2023, myscale(mdy(1,1,@)) local(ticks)
18993 19359 19724 20089 20454 20820 21185 21550 21915 22281 22646 23011

. 
. line foo date, xla(`labels', noticks) xtic(`ticks', tlength(*3)) xtitle("")

The idea is that mylabels and myticks write the code you need -- and insist that it is assigned to local macros so that you don't need to type it out again (or even copy and paste). You may want to tune the ticks and labels with other standard suboptions, as also surveyed in https://journals.sagepub.com/doi/pdf...36867X19874264

After a gestation lasting 19 years. those commands were finally written up in Stata Journal 22(4) and the paper is visible at https://journals.sagepub.com/doi/10....6867X221141058

[Juan Carlos]

Nick, you are a genius.

That's just what I wanted to do. Thanks so much for the recommendations. I think I still have a ways to go (I will never finish learning about this) to read a lot more on the subject.

Thank you very much.

Greetings!!!