R help - Jan 2008 - recoding one variable into another - but differently for different cases

Hello,
I have 2 variables in my sample Data: Data$A and Data$B
Variable Data$A can assume values: 1, 2, 3, and 4.
Variable Data$B identifies my cases and can assume values: 1 and 2.

I need to recode my variable Data$A into a new variable Data$new such that:

People who are Data[Data$B %in% 1, ]  are recoded like this:

Value on Data$A    Value on Data$new
1                             +1
2                             -1
3                             0
4                             99

People who are Data[Data$B %in% 2, ]  are recoded like this:

Value on Data$A    Value on Data$new
1                             -1
2                             +1
3                             0
4                             99

I am having hard time doing this. Any help would be greatly appreciated.
Dimitri

Dimitri Liakhovitski wrote:
> Hello,
> I have 2 variables in my sample Data: Data$A and Data$B
> Variable Data$A can assume values: 1, 2, 3, and 4.
> Variable Data$B identifies my cases and can assume values: 1 and 2.
> 
> I need to recode my variable Data$A into a new variable Data$new such that:
> 
> People who are Data[Data$B %in% 1, ]  are recoded like this:
> 
> Value on Data$A    Value on Data$new
> 1                             +1
> 2                             -1
> 3                             0
> 4                             99
> 
> People who are Data[Data$B %in% 2, ]  are recoded like this:
> 
> Value on Data$A    Value on Data$new
> 1                             -1
> 2                             +1
> 3                             0
> 4                             99
> 
> I am having hard time doing this. Any help would be greatly appreciated.
> Dimitri
How about this:
> Data
   A B
14 2 2
12 4 2
6  2 1
10 2 2
15 3 2
9  1 2
8  4 1
3  3 1
4  4 1
11 3 2
16 4 2
5  1 1
2  2 1
7  3 1
13 1 2
1  1 1


# Create a vector of the codes, in sequence for each subset
Codes <- c("+1", "-1", "0", "99",
"-1", "+1", "0", "99")

# Create 'new' using indices into 'Codes'
Data$new <- with(Data, Codes[((B - 1) * 4) + A])
> Data
   A B new
14 2 2  +1
12 4 2  99
6  2 1  -1
10 2 2  +1
15 3 2   0
9  1 2  -1
8  4 1  99
3  3 1   0
4  4 1  99
11 3 2   0
16 4 2  99
5  1 1  +1
2  2 1  -1
7  3 1   0
13 1 2  -1
1  1 1  +1



HTH,

Marc Schwartz

You could create a lookup table or use recode in the car package.

Another possibility is to use a logical/arithmetic expression.  The
following expression says that

- if A is 1 then use the first term equals the coefficient, namely 1
if B ==1 and -1 if B == 2.
  Also, if A is not 1 then that term is zero and can be ignored.
- if A is 2 or 99 then the second or third terms are used analogously
- otherwise no terms are selected and the expression equals zero

   transform(Data, new       (A == 1) * ((B == 1) - (B == 2)) + (A == 2) * ((B
== 2) - (B =1)) + (A == 4) * 99)

This could be reduced even more although at the expense of
understandability, e.g.

   transform(Data, new = ifelse(A > 2, 99 * (A == 4), (A == B) - (A != B)))

On Jan 22, 2008 12:25 PM, Dimitri Liakhovitski <ld7631 at gmail.com>
wrote:
> Hello,
> I have 2 variables in my sample Data: Data$A and Data$B
> Variable Data$A can assume values: 1, 2, 3, and 4.
> Variable Data$B identifies my cases and can assume values: 1 and 2.
>
> I need to recode my variable Data$A into a new variable Data$new such that:
>
> People who are Data[Data$B %in% 1, ]  are recoded like this:
>
> Value on Data$A    Value on Data$new
> 1                             +1
> 2                             -1
> 3                             0
> 4                             99
>
> People who are Data[Data$B %in% 2, ]  are recoded like this:
>
> Value on Data$A    Value on Data$new
> 1                             -1
> 2                             +1
> 3                             0
> 4                             99
>
> I am having hard time doing this. Any help would be greatly appreciated.
> Dimitri
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

Slight correction of the English:

- if A is 1 then the first term equals the coefficient of (A == 1).
  That is the first term equals 1 if B==1 and equals -1 if B==2.
  If A does not equal 1 then the first term is zero and can be ignored.
- terms 2 and 3 are interpreted analogously
- if A==3 (or other value) then the (A==?) part of each term equals
  zero so the entire expression is zero.

On Jan 22, 2008 2:08 PM, Gabor Grothendieck <ggrothendieck at gmail.com>
wrote:
> You could create a lookup table or use recode in the car package.
>
> Another possibility is to use a logical/arithmetic expression.  The
> following expression says that
>
> - if A is 1 then use the first term equals the coefficient, namely 1
> if B ==1 and -1 if B == 2.
>  Also, if A is not 1 then that term is zero and can be ignored.
> - if A is 2 or 99 then the second or third terms are used analogously
> - otherwise no terms are selected and the expression equals zero
>
>   transform(Data, new >      (A == 1) * ((B == 1) - (B == 2)) + (A == 2)
* ((B == 2) - (B => 1)) + (A == 4) * 99)
>
> This could be reduced even more although at the expense of
> understandability, e.g.
>
>   transform(Data, new = ifelse(A > 2, 99 * (A == 4), (A == B) - (A !=
B)))
>
>
> On Jan 22, 2008 12:25 PM, Dimitri Liakhovitski <ld7631 at gmail.com>
wrote:
> > Hello,
> > I have 2 variables in my sample Data: Data$A and Data$B
> > Variable Data$A can assume values: 1, 2, 3, and 4.
> > Variable Data$B identifies my cases and can assume values: 1 and 2.
> >
> > I need to recode my variable Data$A into a new variable Data$new such
that:
> >
> > People who are Data[Data$B %in% 1, ]  are recoded like this:
> >
> > Value on Data$A    Value on Data$new
> > 1                             +1
> > 2                             -1
> > 3                             0
> > 4                             99
> >
> > People who are Data[Data$B %in% 2, ]  are recoded like this:
> >
> > Value on Data$A    Value on Data$new
> > 1                             -1
> > 2                             +1
> > 3                             0
> > 4                             99
> >
> > I am having hard time doing this. Any help would be greatly
appreciated.
> > Dimitri
> >
> > ______________________________________________
> > R-help at r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
> >
>

Thanks a lot, everyone!
Dimitri

On 1/22/08, Gabor Grothendieck <ggrothendieck at gmail.com>
wrote:
> Slight correction of the English:
>
> - if A is 1 then the first term equals the coefficient of (A == 1).
>  That is the first term equals 1 if B==1 and equals -1 if B==2.
>  If A does not equal 1 then the first term is zero and can be ignored.
> - terms 2 and 3 are interpreted analogously
> - if A==3 (or other value) then the (A==?) part of each term equals
>  zero so the entire expression is zero.
>
> On Jan 22, 2008 2:08 PM, Gabor Grothendieck <ggrothendieck at
gmail.com> wrote:
> > You could create a lookup table or use recode in the car package.
> >
> > Another possibility is to use a logical/arithmetic expression.  The
> > following expression says that
> >
> > - if A is 1 then use the first term equals the coefficient, namely 1
> > if B ==1 and -1 if B == 2.
> >  Also, if A is not 1 then that term is zero and can be ignored.
> > - if A is 2 or 99 then the second or third terms are used analogously
> > - otherwise no terms are selected and the expression equals zero
> >
> >   transform(Data, new > >      (A == 1) * ((B == 1) - (B == 2))
+ (A == 2) * ((B == 2) - (B => > 1)) + (A == 4) * 99)
> >
> > This could be reduced even more although at the expense of
> > understandability, e.g.
> >
> >   transform(Data, new = ifelse(A > 2, 99 * (A == 4), (A == B) - (A
!= B)))
> >
> >
> > On Jan 22, 2008 12:25 PM, Dimitri Liakhovitski <ld7631 at
gmail.com> wrote:
> > > Hello,
> > > I have 2 variables in my sample Data: Data$A and Data$B
> > > Variable Data$A can assume values: 1, 2, 3, and 4.
> > > Variable Data$B identifies my cases and can assume values: 1 and
2.
> > >
> > > I need to recode my variable Data$A into a new variable Data$new
such that:
> > >
> > > People who are Data[Data$B %in% 1, ]  are recoded like this:
> > >
> > > Value on Data$A    Value on Data$new
> > > 1                             +1
> > > 2                             -1
> > > 3                             0
> > > 4                             99
> > >
> > > People who are Data[Data$B %in% 2, ]  are recoded like this:
> > >
> > > Value on Data$A    Value on Data$new
> > > 1                             -1
> > > 2                             +1
> > > 3                             0
> > > 4                             99
> > >
> > > I am having hard time doing this. Any help would be greatly
appreciated.
> > > Dimitri
> > >
> > > ______________________________________________
> > > R-help at r-project.org mailing list
> > > https://stat.ethz.ch/mailman/listinfo/r-help
> > > PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> > > and provide commented, minimal, self-contained, reproducible
code.
> > >
> >
>

No one else mentioned this, but if those 99s represent missings, you
should be using NA not a special numeric value.

Hadley

On Jan 22, 2008 5:40 PM, Dimitri Liakhovitski <ld7631 at gmail.com>
wrote:
> Thanks a lot, everyone!
> Dimitri
>
>
> On 1/22/08, Gabor Grothendieck <ggrothendieck at gmail.com> wrote:
> > Slight correction of the English:
> >
> > - if A is 1 then the first term equals the coefficient of (A == 1).
> >  That is the first term equals 1 if B==1 and equals -1 if B==2.
> >  If A does not equal 1 then the first term is zero and can be ignored.
> > - terms 2 and 3 are interpreted analogously
> > - if A==3 (or other value) then the (A==?) part of each term equals
> >  zero so the entire expression is zero.
> >
> > On Jan 22, 2008 2:08 PM, Gabor Grothendieck <ggrothendieck at
gmail.com> wrote:
> > > You could create a lookup table or use recode in the car package.
> > >
> > > Another possibility is to use a logical/arithmetic expression. 
The
> > > following expression says that
> > >
> > > - if A is 1 then use the first term equals the coefficient,
namely 1
> > > if B ==1 and -1 if B == 2.
> > >  Also, if A is not 1 then that term is zero and can be ignored.
> > > - if A is 2 or 99 then the second or third terms are used
analogously
> > > - otherwise no terms are selected and the expression equals zero
> > >
> > >   transform(Data, new > > >      (A == 1) * ((B == 1) -
(B == 2)) + (A == 2) * ((B == 2) - (B => > > 1)) + (A == 4) * 99)
> > >
> > > This could be reduced even more although at the expense of
> > > understandability, e.g.
> > >
> > >   transform(Data, new = ifelse(A > 2, 99 * (A == 4), (A == B)
- (A != B)))
> > >
> > >
> > > On Jan 22, 2008 12:25 PM, Dimitri Liakhovitski <ld7631 at
gmail.com> wrote:
> > > > Hello,
> > > > I have 2 variables in my sample Data: Data$A and Data$B
> > > > Variable Data$A can assume values: 1, 2, 3, and 4.
> > > > Variable Data$B identifies my cases and can assume values: 1
and 2.
> > > >
> > > > I need to recode my variable Data$A into a new variable
Data$new such that:
> > > >
> > > > People who are Data[Data$B %in% 1, ]  are recoded like this:
> > > >
> > > > Value on Data$A    Value on Data$new
> > > > 1                             +1
> > > > 2                             -1
> > > > 3                             0
> > > > 4                             99
> > > >
> > > > People who are Data[Data$B %in% 2, ]  are recoded like this:
> > > >
> > > > Value on Data$A    Value on Data$new
> > > > 1                             -1
> > > > 2                             +1
> > > > 3                             0
> > > > 4                             99
> > > >
> > > > I am having hard time doing this. Any help would be greatly
appreciated.
> > > > Dimitri
> > > >
> > > > ______________________________________________
> > > > R-help at r-project.org mailing list
> > > > https://stat.ethz.ch/mailman/listinfo/r-help
> > > > PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> > > > and provide commented, minimal, self-contained, reproducible
code.
> > > >
> > >
> >
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>


-- 
http://had.co.nz/

Note that if you do use NA rather than 99 as others have suggested then
the A==4 term should use ifelse rather than multiplication since 0 *
NA = NA, not 0:

transform(Data, new    (A == 1) * ((B == 1) - (B == 2)) + (A == 2) * ((B == 2) -
(B ==1))
+ ifelse(A == 4, NA, 0))

In fact, although more verbose, if you find ifelse clearer you might
use it in all three terms:

On Jan 22, 2008 5:01 PM, Gabor Grothendieck <ggrothendieck at gmail.com>
wrote:
> Slight correction of the English:
>
> - if A is 1 then the first term equals the coefficient of (A == 1).
>  That is the first term equals 1 if B==1 and equals -1 if B==2.
>  If A does not equal 1 then the first term is zero and can be ignored.
> - terms 2 and 3 are interpreted analogously
> - if A==3 (or other value) then the (A==?) part of each term equals
>  zero so the entire expression is zero.
>
>
> On Jan 22, 2008 2:08 PM, Gabor Grothendieck <ggrothendieck at
gmail.com> wrote:
> > You could create a lookup table or use recode in the car package.
> >
> > Another possibility is to use a logical/arithmetic expression.  The
> > following expression says that
> >
> > - if A is 1 then use the first term equals the coefficient, namely 1
> > if B ==1 and -1 if B == 2.
> >  Also, if A is not 1 then that term is zero and can be ignored.
> > - if A is 2 or 99 then the second or third terms are used analogously
> > - otherwise no terms are selected and the expression equals zero
> >
> >   transform(Data, new > >      (A == 1) * ((B == 1) - (B == 2))
+ (A == 2) * ((B == 2) - (B => > 1)) + (A == 4) * 99)
> >
> > This could be reduced even more although at the expense of
> > understandability, e.g.
> >
> >   transform(Data, new = ifelse(A > 2, 99 * (A == 4), (A == B) - (A
!= B)))
> >
> >
> > On Jan 22, 2008 12:25 PM, Dimitri Liakhovitski <ld7631 at
gmail.com> wrote:
> > > Hello,
> > > I have 2 variables in my sample Data: Data$A and Data$B
> > > Variable Data$A can assume values: 1, 2, 3, and 4.
> > > Variable Data$B identifies my cases and can assume values: 1 and
2.
> > >
> > > I need to recode my variable Data$A into a new variable Data$new
such that:
> > >
> > > People who are Data[Data$B %in% 1, ]  are recoded like this:
> > >
> > > Value on Data$A    Value on Data$new
> > > 1                             +1
> > > 2                             -1
> > > 3                             0
> > > 4                             99
> > >
> > > People who are Data[Data$B %in% 2, ]  are recoded like this:
> > >
> > > Value on Data$A    Value on Data$new
> > > 1                             -1
> > > 2                             +1
> > > 3                             0
> > > 4                             99
> > >
> > > I am having hard time doing this. Any help would be greatly
appreciated.
> > > Dimitri
> > >
> > > ______________________________________________
> > > R-help at r-project.org mailing list
> > > https://stat.ethz.ch/mailman/listinfo/r-help
> > > PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> > > and provide commented, minimal, self-contained, reproducible
code.
> > >
> >
>