R help - Nov 2008 - how can I save the estimates of a regression model in a file?

Dear all
I need some help with R.
How can I save the estimates of a regression model in a file?

here is what I did:

1) this is my regression model:
fit1 <- lm(logmilk ~ logdays + days, data=data2)

2) however, I want to get the parameters estimates for each individual (by
group):
so i did the following:
by(data2, list(data2$V2),function(.data2) lm(logmilk ~ logdays + days,
data.data2))

3) Then to keep the estimates in a file I did:
res1 <- by(data2, list(data2$V2),function(.data2) lm(logmilk ~ logdays + V5,
data= .data2))

4) and this is what I  get for each individual:
: 1
Call:
lm(formula = logmilk ~ logdays + days, data = .data2)
Coefficients:
(Intercept)      logdays           days
   3.414105     0.069387    -0.001732

---------------------------------------------------------------------------
: 2
Call:
lm(formula = logmilk ~ logdays + days, data = .data2)

Coefficients:
(Intercept)      logdays           days
  3.2276114    0.1223412   -0.0006836

and so on:


5) THE QUESTION is:

there is a way in R to get an output file as you get in SAS when you use:
prog reg data=xxx outest=estimate;
I would need an output that looks like this:

individual     intercept     logdays   days           etc
1           3.414105        0.069387   -0.0006836
2           3.2276114       0.1223412  -0.0006836

n           ........               ..........       .........
THANKS FOR YOUR HELP

Maria

-- 
-----------------------------
pilar.schneider@gmail.com

	[[alternative HTML version deleted]]

On 11/5/2008 12:12 PM, pilar schneider wrote:
> Dear all
> I need some help with R.
> How can I save the estimates of a regression model in a file?
> 
> here is what I did:
> 
> 1) this is my regression model:
> fit1 <- lm(logmilk ~ logdays + days, data=data2)
> 
> 2) however, I want to get the parameters estimates for each individual (by
> group):
> so i did the following:
> by(data2, list(data2$V2),function(.data2) lm(logmilk ~ logdays + days,
data> .data2))
> 
> 3) Then to keep the estimates in a file I did:
> res1 <- by(data2, list(data2$V2),function(.data2) lm(logmilk ~ logdays +
V5,
> data= .data2))
> 
> 4) and this is what I  get for each individual:
> : 1
> Call:
> lm(formula = logmilk ~ logdays + days, data = .data2)
> Coefficients:
> (Intercept)      logdays           days
>    3.414105     0.069387    -0.001732
> 
> ---------------------------------------------------------------------------
> : 2
> Call:
> lm(formula = logmilk ~ logdays + days, data = .data2)
> 
> Coefficients:
> (Intercept)      logdays           days
>   3.2276114    0.1223412   -0.0006836
> 
> and so on:
> 
> 
> 5) THE QUESTION is:
> 
> there is a way in R to get an output file as you get in SAS when you use:
> prog reg data=xxx outest=estimate;
> I would need an output that looks like this:
> 
> individual     intercept     logdays   days           etc
> 1           3.414105        0.069387   -0.0006836
> 2           3.2276114       0.1223412  -0.0006836
> 
> n           ........               ..........       .........
> THANKS FOR YOUR HELP
> 
> Maria 
  Consider something like this:

as.data.frame(t(sapply(split(data2, data2$V2),
function(x){coef(lm(logmilk ~ logdays + days, data = x))})))

-- 
Chuck Cleland, Ph.D.
NDRI, Inc. (www.ndri.org)
71 West 23rd Street, 8th floor
New York, NY 10010
tel: (212) 845-4495 (Tu, Th)
tel: (732) 512-0171 (M, W, F)
fax: (917) 438-0894

On Wed, Nov 5, 2008 at 11:12 AM, pilar schneider
<pilar.schneider at gmail.com> wrote:
> Dear all
> I need some help with R.
> How can I save the estimates of a regression model in a file?
>
> here is what I did:
>
> 1) this is my regression model:
> fit1 <- lm(logmilk ~ logdays + days, data=data2)
>
> 2) however, I want to get the parameters estimates for each individual (by
> group):
> so i did the following:
> by(data2, list(data2$V2),function(.data2) lm(logmilk ~ logdays + days,
data> .data2))
>
> 3) Then to keep the estimates in a file I did:
> res1 <- by(data2, list(data2$V2),function(.data2) lm(logmilk ~ logdays +
V5,
> data= .data2))
>
Here's one way that uses the plyr package (http://had.co.nz/)

# Name the model fitting function
fit_model <- function(df) lm(logmilk ~ logdays + days, data = df)

# Create a list of models for each value of V2
# this is basically equivalent to by, but it deals with labelling a bit better
models <- dlply(data2, .(V2), fit_model)

# Produce a data frame containing the coefficients for each model
ldply(models, coef)

Hadley


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