R help - Dec 2007 - Help with lm and multiple linear regression? (Plain Text version)

(Apologies the previous version was sent as rich text)

Hello,
I'm new to R, but I've read the intro to R and successfully connected it
to an instance of mysql.  I'm trying to perform multiple linear regression,
but I'm having trouble using the lm function.  To start, I have read in a
simply y matrix of values(dependent variable) and x matrix of independent
variables.  It says both are data frames, but lm is giving me an error that my y
variable is a list.

Any suggestions on how to do this?  It's not clear to me what the problem is
as they're both data frames.  My actual problem will use a much wider matrix
of coefficients, I've only included two for illustration.

Additionally, I'd actually like to weight the observations.  How would I go
about doing that?  I also have that as a separate column vector.

Thanks,
Aaron

Here's my session:
> margin
    margin
1    66.67
2   -58.33
3   100.00
4   -33.33
5   200.00
6   -83.33
7  -100.00
8     0.00
9   100.00
10  -18.18
11  -55.36
12 -125.00
13  -33.33
14 -200.00
15    0.00
16 -100.00
17   75.00
18    0.00
19 -200.00
20   35.71
21  100.00
22   50.00
23  -86.67
24  165.00
> personcoeff
   Person1 Person2
1       -1       1
2       -1       1
3       -1       1
4       -1       1
5       -1       1
6       -1       1
7        0       0
8        0       0
9        0       1
10      -1       1
11      -1       1
12      -1       1
13      -1       1
14      -1       0
15       0       0
16       0       0
17       0       1
18      -1       1
19      -1       1
20      -1       1
21      -1       1
22      -1       1
23      -1       1
24      -1       1
> class(margin)
[1] "data.frame"
> class(personcoeff)
[1] "data.frame"
> lm(margin~personcoeff)
Error in model.frame(formula, rownames, variables, varnames, extras, extranames,
:
        invalid type (list) for variable 'margin'


     
____________________________________________________________________________________
Be a better friend, newshound, and

consider merging everything into a singe dataframe.  i haven't tried
it, but something like the following could work:
> reg.data <- cbind(margin, personcoeff)
> names(reg.data) <- c('margin', 'p1', 'p2')
> lm(margin~p1+p2, data = reg.data)
the idea here is that by specifying the data frame with the data
argument in lm, R looks for the columns of the names specified in the
formula.

for weights, see ?lm and look for the weights argument.

cheers,
tc

On Dec 28, 2007 10:22 AM, Aaron Barzilai <aaron_barzilai at yahoo.com>
wrote:
> (Apologies the previous version was sent as rich text)
>
> Hello,
> I'm new to R, but I've read the intro to R and successfully
connected it to an instance of mysql.  I'm trying to perform multiple linear
regression, but I'm having trouble using the lm function.  To start, I have
read in a simply y matrix of values(dependent variable) and x matrix of
independent variables.  It says both are data frames, but lm is giving me an
error that my y variable is a list.
>
> Any suggestions on how to do this?  It's not clear to me what the
problem is as they're both data frames.  My actual problem will use a much
wider matrix of coefficients, I've only included two for illustration.
>
> Additionally, I'd actually like to weight the observations.  How would
I go about doing that?  I also have that as a separate column vector.
>
> Thanks,
> Aaron
>
> Here's my session:
> > margin
>     margin
> 1    66.67
> 2   -58.33
> 3   100.00
> 4   -33.33
> 5   200.00
> 6   -83.33
> 7  -100.00
> 8     0.00
> 9   100.00
> 10  -18.18
> 11  -55.36
> 12 -125.00
> 13  -33.33
> 14 -200.00
> 15    0.00
> 16 -100.00
> 17   75.00
> 18    0.00
> 19 -200.00
> 20   35.71
> 21  100.00
> 22   50.00
> 23  -86.67
> 24  165.00
> > personcoeff
>    Person1 Person2
> 1       -1       1
> 2       -1       1
> 3       -1       1
> 4       -1       1
> 5       -1       1
> 6       -1       1
> 7        0       0
> 8        0       0
> 9        0       1
> 10      -1       1
> 11      -1       1
> 12      -1       1
> 13      -1       1
> 14      -1       0
> 15       0       0
> 16       0       0
> 17       0       1
> 18      -1       1
> 19      -1       1
> 20      -1       1
> 21      -1       1
> 22      -1       1
> 23      -1       1
> 24      -1       1
> > class(margin)
> [1] "data.frame"
> > class(personcoeff)
> [1] "data.frame"
> > lm(margin~personcoeff)
> Error in model.frame(formula, rownames, variables, varnames, extras,
extranames,  :
>         invalid type (list) for variable 'margin'
>
>
>      
____________________________________________________________________________________
> Be a better friend, newshound, and
>
> ______________________________________________
> 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.
>


-- 
Tim Calkins
0406 753 997

Tim (and others who responded privately),

Thanks for the help, this approach did work.  I have also reread ?lm a little
more closely, I do see the weights functionality.

I have one last question: Now that I understand how to call this function and
review the results, I want to extend it to my much larger real problem, with
100s of columns.  Is there a way to call the function in more of a matrix
algebra syntax, where I would list the matrix(e.g. personcoeff) rather than the
individual column names?  It seems like I might need to use lm.wfit, but per the
help I'd rather use lm.

Thanks,
Aaron




----- Original Message ----
From: Tim Calkins <tim.calkins at gmail.com>
To: Aaron Barzilai <aaron_barzilai at yahoo.com>
Cc: r-help at r-project.org
Sent: Thursday, December 27, 2007 6:55:57 PM
Subject: Re: [R] Help with lm and multiple linear regression? (Plain Text
version)

consider merging everything into a singe dataframe.  i haven't tried
it, but something like the following could work:
> reg.data <- cbind(margin, personcoeff)
> names(reg.data) <- c('margin', 'p1', 'p2')
> lm(margin~p1+p2, data = reg.data)
the idea here is that by specifying the data frame with the data
argument in lm, R looks for the columns of the names specified in the
formula.

for weights, see ?lm and look for the weights argument.

cheers,
tc

On Dec 28, 2007 10:22 AM, Aaron Barzilai <aaron_barzilai at yahoo.com>
wrote:
> (Apologies the previous version was sent as rich text)
>
> Hello,
> I'm new to R, but I've read the intro to R and successfully
connected it to an instance of mysql.  I'm trying to perform multiple linear
regression, but I'm having trouble using the lm function.  To start, I have
read in a simply y matrix of values(dependent variable) and x matrix of
independent variables.  It says both are data frames, but lm is giving me an
error that my y variable is a list.
>
> Any suggestions on how to do this?  It's not clear to me what the
problem is as they're both data frames.  My actual problem will use a much
wider matrix of coefficients, I've only included two for illustration.
>
> Additionally, I'd actually like to weight the observations.  How would
I go about doing that?  I also have that as a separate column vector.
>
> Thanks,
> Aaron
>
> Here's my session:
> > margin
>    margin
> 1    66.67
> 2  -58.33
> 3  100.00
> 4  -33.33
> 5  200.00
> 6  -83.33
> 7  -100.00
> 8    0.00
> 9  100.00
> 10  -18.18
> 11  -55.36
> 12 -125.00
> 13  -33.33
> 14 -200.00
> 15    0.00
> 16 -100.00
> 17  75.00
> 18    0.00
> 19 -200.00
> 20  35.71
> 21  100.00
> 22  50.00
> 23  -86.67
> 24  165.00
> > personcoeff
>    Person1 Person2
> 1      -1      1
> 2      -1      1
> 3      -1      1
> 4      -1      1
> 5      -1      1
> 6      -1      1
> 7        0      0
> 8        0      0
> 9        0      1
> 10      -1      1
> 11      -1      1
> 12      -1      1
> 13      -1      1
> 14      -1      0
> 15      0      0
> 16      0      0
> 17      0      1
> 18      -1      1
> 19      -1      1
> 20      -1      1
> 21      -1      1
> 22      -1      1
> 23      -1      1
> 24      -1      1
> > class(margin)
> [1] "data.frame"
> > class(personcoeff)
> [1] "data.frame"
> > lm(margin~personcoeff)
> Error in model.frame(formula, rownames, variables, varnames, extras,
extranames,  :
>        invalid type (list) for variable 'margin'
>
>
>     
____________________________________________________________________________________
> Be a better friend, newshound, and
>
> ______________________________________________
> 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.
>


-- 
Tim Calkins
0406 753 997


     
____________________________________________________________________________________
Be a better friend, newshound, and