[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Replace

From	Nick Winter <[email protected]>
To	[email protected]
Subject	Re: st: Replace
Date	Fri, 21 Nov 2003 15:15:38 -0500

Hi,

As I understand it, aside from the "PC" complication, your task is to propagate values of status forward through the data for each country. This can be accomplished by:

. bysort country (date1): replace status=status[_n-1] ///
if status=="" & inlist(status[_n-1],"M","W","NM")

What this does:

. bysort country (date1):

executes the command within country, with cases sorted by date1 within country.

replace status=status[_n-1]

replaces status with the value of status in the prior observation, but only...

if status=="" & inlist(status[_n-1],"M","W","NM")

if status is blank, and the prior observation for status is M, W, or NM. Because Stata executes the -replace- from top to bottom, each blank will be replaced in order, and the value of M/W/MN will propagate through the observations. Because of teh -bysort- prefix, values will not propagate into the next country.

But this leaves you with the "PC" issue: with observations of "PC", you want to look *forward* to the next non-blank, grab that, then continue as above.

Here's my take on that:

. bysort country (date1): replace status = "X" ///
if status=="" & inlist(status[_n-1],"PC","X")

. gen negdate = -date1
. bysort country (negdate): replace status = status[_n-1] ///
if (status=="X" | status=="PC") inlist(status[_n-1],"M","W","NM")

. bysort country (date1): replace status=status[_n-1] ///
if (status=="" | & inlist(status[_n-1],"M","W","NM")

. replace status="" if status=="X"
. drop negdate

What this does:

First, fill in with "X" all the blank observations that follow "PC" (logic here is the same as the propagation above).

Then, reverse the date sort order (by sorting on the negative of the date variable), and propagate M/W/NM through the "X" values (and the PC values).

Then, do the "simple" propagation, as above.

Then, delete any remaining "X" values. (There will be remaining "X" values in any cases where a country has a "PC" that is *not* followed ever by a M/W/NM at any point.)

-Nick Winter

At 02:09 PM 11/21/2003 -0500, you wrote:

Dear Statapeeps,

I need to replace some values of a variable that meet certain
conditions. Although I know it's not possible to make direct references
to rows using the -replace- command, this is what I want to do:

replace status = status[`j'] if status[`j'] == "NM" | status[`j'] == "W"
in `i'

I tried to use a local variable to avoid using status[`j'] in the
command line but it didn't work. Like this:

                                        local sta = status[`j']
                                        replace status = "`sta'" if
status[`j']=="NM"|status[`j']=="W" in `i'

You will find my whole script below.

Do you know how I can get past this problem?

Here I present a chunk of my dataset so that you understand the problem
better. You don't need to read the rest if you already understood what I
want to do from the command line above.

The data is organized in a panel format by country-date. What I need is
to fill in the values of 'status' whenever they are missing and they
meet certain conditions.

Suppose I start a loop that sweeps through 'status' row by row. Most
observations in 'status' are missing and suddenly, in observation [i] I
find one cell that is not missing.
(a) If status[i] = "M"/"W"/"NM", the subsequent rows (observations [j],
where j gets different values) should be "M"/"W"/"NM" whenever
country[i]=country[j] and date1[i]=date1[j].

(b) If, however, status[i] = "PC" then I need to look at the next
non-empty cell to decide what "PC" is going to be in [i]. PC is going to
be converted into "M" except if a subsequent value is "W" or "NM". A
value in row [j] is considered subsequent if [j] > [i] and
country[i]=country[j] and date1[i]=date1[j].

The data looks like this:

               country    date      date1    date2     status
  401. |       Armenia    1997m5    1995m6   1999m12          |
  402. |       Armenia    1997m6    1996m2   1999m12       PC |
  403. |       Armenia    1997m7    1996m2   1999m12          |
  404. |       Armenia    1997m8    1996m2   1999m12          |
  405. |       Armenia    1997m9    1996m2   1999m12          |
       |------------------------------------------------------|
  406. |       Armenia   1997m10    1996m2   1999m12          |
  407. |       Armenia   1997m11    1996m2   1999m12          |
  408. |       Armenia   1997m12    1996m2   1999m12          |
  409. |       Armenia    1998m1    1996m2   1999m12          |
  410. |       Armenia    1998m2    1996m2   1999m12        W |
       |------------------------------------------------------|
  411. |       Armenia    1998m3    1996m2   1999m12          |
  412. |       Armenia    1998m4    1996m2   1999m12          |
  413. |       Armenia    1998m5    1996m2   1999m12          |
  414. |       Armenia    1998m6    1996m2   1999m12          |
  415. |       Armenia    1998m7    1996m2   1999m12          |
       |------------------------------------------------------|
  416. |       Armenia    1998m8    1996m2   1999m12          |
  417. |       Armenia    1998m9    1996m2   1999m12          |
  418. |       Armenia   1998m10    1996m2   1999m12          |
  419. |       Armenia   1998m11    1996m2   1999m12          |
  420. |       Armenia   1998m12    1996m2   1999m12       PC |
       |------------------------------------------------------|
  421. |       Armenia    1999m1    1996m2   1999m12          |
  422. |       Armenia    1999m2    1996m2   1999m12          |
  423. |       Armenia    1999m3    1996m2   1999m12          |
  424. |       Armenia    1999m4    1996m2   1999m12          |
  425. |       Armenia    1999m5    1996m2   1999m12          |
       |------------------------------------------------------|
  426. |       Armenia    1999m6    1996m2   1999m12          |
  427. |       Armenia    1999m7    1996m2   1999m12          |
  428. |       Armenia    1999m8    1996m2   1999m12          |
  429. |       Armenia    1999m9    1996m2   1999m12          |
  430. |       Armenia   1999m10    1996m2   1999m12        M |
       |------------------------------------------------------|
  431. |       Armenia   1999m11    1996m2   1999m12          |
  432. |       Armenia   1999m12    1996m2   1999m12          |
  433. |       Armenia    2000m1         .         .          |
  434. |       Armenia    2000m2         .         .          |

And after running my script it should look like this:

               country    date      date1    date2     status
  401. |       Armenia    1997m5    1995m6   1999m12          |
  402. |       Armenia    1997m6    1996m2   1999m12        W |
  403. |       Armenia    1997m7    1996m2   1999m12        W |
  404. |       Armenia    1997m8    1996m2   1999m12        W |
  405. |       Armenia    1997m9    1996m2   1999m12        W |
       |------------------------------------------------------|
  406. |       Armenia   1997m10    1996m2   1999m12        W |
  407. |       Armenia   1997m11    1996m2   1999m12        W |
  408. |       Armenia   1997m12    1996m2   1999m12        W |
  409. |       Armenia    1998m1    1996m2   1999m12        W |
  410. |       Armenia    1998m2    1996m2   1999m12        W |
       |------------------------------------------------------|
  411. |       Armenia    1998m3    1996m2   1999m12        W |
  412. |       Armenia    1998m4    1996m2   1999m12        W |
  413. |       Armenia    1998m5    1996m2   1999m12        W |
  414. |       Armenia    1998m6    1996m2   1999m12        W |
  415. |       Armenia    1998m7    1996m2   1999m12        W |
       |------------------------------------------------------|
  416. |       Armenia    1998m8    1996m2   1999m12        W |
  417. |       Armenia    1998m9    1996m2   1999m12        W |
  418. |       Armenia   1998m10    1996m2   1999m12        W |
  419. |       Armenia   1998m11    1996m2   1999m12        W |
  420. |       Armenia   1998m12    1996m2   1999m12        M |
       |------------------------------------------------------|
  421. |       Armenia    1999m1    1996m2   1999m12        M |
  422. |       Armenia    1999m2    1996m2   1999m12        M |
  423. |       Armenia    1999m3    1996m2   1999m12        M |
  424. |       Armenia    1999m4    1996m2   1999m12        M |
  425. |       Armenia    1999m5    1996m2   1999m12        M |
       |------------------------------------------------------|
  426. |       Armenia    1999m6    1996m2   1999m12        M |
  427. |       Armenia    1999m7    1996m2   1999m12        M |
  428. |       Armenia    1999m8    1996m2   1999m12        M |
  429. |       Armenia    1999m9    1996m2   1999m12        M |
  430. |       Armenia   1999m10    1996m2   1999m12        M |
       |------------------------------------------------------|
  431. |       Armenia   1999m11    1996m2   1999m12        M |
  432. |       Armenia   1999m12    1996m2   1999m12        M |
  433. |       Armenia    2000m1         .         .          |
  434. |       Armenia    2000m2         .         .          |

This is my script and I don't know what the problem is:

      local m = _N
        forv i = 1/`m' {
                if status[`i'] != "" {
                        if status[`i'] == "PC" {
                                local j = `i'+1
                                while status[`j'] == "" &
country[`j']==country[`i'] & date1[`i']==date1[`j'] {
                                        local sta = status[`j']
                                        replace status = "`sta'" if
status[`j']=="NM"|status[`j']=="W" in `i'
                                        replace status = "M" if
status[`j']=="M"|(date1[`j']!=date1[`i'] & country[`j']!=country[`i'] in
`i'
                                }
                        }
                        else {
                                local j = `i'+1
                                local k = 0
                                while status[`j'] == "" &
country[`i']==country[`j'] & date1[`i']==date1[`j'] {
                                        local sta = status[`i']
                                        replace status = "`sta'" in `j'
                                        local j = `j'+1
                                        local k = `k'+1
                                }
                        }
                        local i = `i'+`k'
                }
        }

Thanks a lot,
Adrian

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

--------------------------------------------------------
Nicholas Winter t 607.255.8819
Department of Government f 607.255.4530
308 White Hall [email protected]
Cornell University falcon.arts.cornell.edu/nw53
Ithaca, NY 14853-4601
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/

References:
- st: Replace
  - From: "de la Garza, Adrian" <[email protected]>

Prev by Date: st: RE: Invalid syntax
Next by Date: st: EM Algorithm
Previous by thread: st: RE: Replace
Next by thread: st: EM Algorithm
Index(es):
- Date
- Thread