Re: st: Creating a series with two observations for each month

[Nick Cox <njcoxstata@gmail.com>]

The only way I know to do this satisfactorily is to -tsset- in terms
of some sequence number and have a separate date variable for plotting
etc. Clearly you are working with a fiction that the dates are equally
spaced.

Here is a token example.

. clear

. set obs 12
obs was 0, now 12

. gen month = _n

. gen year = 2012

. expand 2
(12 observations created)

. sort month

. gen day = cond(mod(_n, 2), 1, 15)

. gen date = mdy(month, day, year)

. format date %td

. gen seq = _n

. tsset seq
        time variable:  seq, 1 to 24
                delta:  1 unit
. list

     +--------------------------------------+
     | month   year   day        date   seq |
     |--------------------------------------|
  1. |     1   2012     1   01jan2012     1 |
  2. |     1   2012    15   15jan2012     2 |
  3. |     2   2012     1   01feb2012     3 |
  4. |     2   2012    15   15feb2012     4 |
  5. |     3   2012     1   01mar2012     5 |
     |--------------------------------------|
  6. |     3   2012    15   15mar2012     6 |
  7. |     4   2012     1   01apr2012     7 |
  8. |     4   2012    15   15apr2012     8 |
  9. |     5   2012     1   01may2012     9 |
 10. |     5   2012    15   15may2012    10 |
     |--------------------------------------|
 11. |     6   2012     1   01jun2012    11 |
 12. |     6   2012    15   15jun2012    12 |
 13. |     7   2012     1   01jul2012    13 |
 14. |     7   2012    15   15jul2012    14 |
 15. |     8   2012     1   01aug2012    15 |
     |--------------------------------------|
 16. |     8   2012    15   15aug2012    16 |
 17. |     9   2012     1   01sep2012    17 |
 18. |     9   2012    15   15sep2012    18 |
 19. |    10   2012     1   01oct2012    19 |
 20. |    10   2012    15   15oct2012    20 |
     |--------------------------------------|
 21. |    11   2012     1   01nov2012    21 |
 22. |    11   2012    15   15nov2012    22 |
 23. |    12   2012     1   01dec2012    23 |
 24. |    12   2012    15   15dec2012    24 |
     +--------------------------------------+
Nick
njcoxstata@gmail.com


On 4 November 2013 00:34, Pedro Nakashima <nakashimapedrolist@gmail.com> wrote:
> Dear statalisters
>
> I would like to construct a time series which is based in the following pattern:
>
> For every month, there are two observations, relative to days 1 and 15.
>
> After that, I would like to plot this series.
>
> I've tried creating date format together with "tsset td, delta(15)",
> but due to month length differences, this combination is creating gaps
> in the series.
>
> Does anyone know how to solve this problem?
>
> Thank you very much,
> Pedro Nakashima.
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/