R help - Aug 2006 - generating an expression for a formula automatically

Hi!

I would like to be able to create formulas automatically. For example, I 
want to be able to create a function that takes on two values: resp and 
x, and then creates the proper formula to regress resp on x.

My code:

fit.main <- function(resp,x) {
 form <- expression(paste(resp," ~
",paste(x,sep="",collapse=" + "),sep=""))
  z <- lm(eval(form))
 z
}
main <-
fit.main("y",c("x1","x2","x3","x4"))

and I get this error:
Error in terms.default(formula, data = data) :
        no terms component

Any suggestions?

Thanks, Maria

Use as.formula to convert the character string to an object of class
"formula"
and note that we want to set the formula's environment appropriately:

fit.main <- function(resp, x, env = parent.frame()) {
	fo <- as.formula(paste("y", "~", paste(x, collapse =
"+")))
	environment(fo) <- env
	fo
}

# test
fit.main("y", letters[1:3])


On 8/24/06, Maria Montez <montez at bu.edu> wrote:
> Hi!
>
> I would like to be able to create formulas automatically. For example, I
> want to be able to create a function that takes on two values: resp and
> x, and then creates the proper formula to regress resp on x.
>
> My code:
>
> fit.main <- function(resp,x) {
>  form <- expression(paste(resp," ~
",paste(x,sep="",collapse=" + "),sep=""))
>  z <- lm(eval(form))
>  z
> }
> main <-
fit.main("y",c("x1","x2","x3","x4"))
>
> and I get this error:
> Error in terms.default(formula, data = data) :
>        no terms component
>
> Any suggestions?
>
> Thanks, Maria
>
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
>

On Thu, 2006-08-24 at 14:01 -0700, Maria Montez wrote:
> Hi!
> 
> I would like to be able to create formulas automatically. For example, I 
> want to be able to create a function that takes on two values: resp and 
> x, and then creates the proper formula to regress resp on x.
> 
> My code:
> 
> fit.main <- function(resp,x) {
>  form <- expression(paste(resp," ~
",paste(x,sep="",collapse=" + "),sep=""))
>   z <- lm(eval(form))
>  z
> }
> main <-
fit.main("y",c("x1","x2","x3","x4"))
> 
> and I get this error:
> Error in terms.default(formula, data = data) :
>         no terms component
> 
> Any suggestions?
> 
> Thanks, Maria
See the last example in ?as.formula:

BTW, I would pay note to the ability to use subset()'s of data frames in
model functions. For example, let's say that your data frame above is
called DF and contains columns 'y' and then 'x1' through
'x50' in
sequence. However, you only want to use the columns you have indicated
in your code above.  You can then do:

  lm(y ~ ., data = subset(DF, select = y:x4))

The use of the '.' on the RHS of the formula indicates to use all other
columns besides the response column in the formula.  In the subset()
function, you can specify a sequential group of columns using the ':'
operator.

For a specific example, let's use the iris data set, which has columns:
> names(iris)
[1] "Sepal.Length" "Sepal.Width"  "Petal.Length"
"Petal.Width"
[5] "Species"

We want to use 'Sepal.Length' as the response variable and then all
columns, other than 'Species', as terms:
> lm(Sepal.Length ~ ., data = subset(iris, select = -Species))
Call:
lm(formula = Sepal.Length ~ ., data = subset(iris, select = -Species))

Coefficients:
 (Intercept)   Sepal.Width  Petal.Length   Petal.Width
      1.8560        0.6508        0.7091       -0.5565


In this case, I excluded the Species columns by using the '-' before the
column name.  However, I could have easily used:
> lm(Sepal.Length ~ ., 
     data = subset(iris, select = Sepal.Length:Petal.Width))

Call:
lm(formula = Sepal.Length ~ ., data = subset(iris, select
Sepal.Length:Petal.Width))

Coefficients:
 (Intercept)   Sepal.Width  Petal.Length   Petal.Width
      1.8560        0.6508        0.7091       -0.5565



See ?subset for additional information.

HTH,

Marc Schwartz

Maria Montez wrote:
> Hi!
> 
> I would like to be able to create formulas automatically. For example, I 
> want to be able to create a function that takes on two values: resp and 
> x, and then creates the proper formula to regress resp on x.
> 
> My code:
> 
> fit.main <- function(resp,x) {
>  form <- expression(paste(resp," ~
",paste(x,sep="",collapse=" + "),sep=""))
>   z <- lm(eval(form))
>  z
> }
> main <-
fit.main("y",c("x1","x2","x3","x4"))
> 
> and I get this error:
> Error in terms.default(formula, data = data) :
>         no terms component
> 
> Any suggestions?
> 
> Thanks, Maria
> 
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
Hi, Maria,

Try

regr <- paste(x, collapse = "+")
form <- as.formula(sprintf("%s ~ %s", resp, regr))

HTH,

--sundar