[R] Pass Conditional & loop argument into a function

[S Ellison]

First thing; are you trying to fit a model specified as

y ~ X + X^2 + X^3 ?

... because if you are you're unlikely to get anything useful. That uses formula syntax in which ^ does not have the arithmetic power meaning; see section 11.1 'Defining statistical models; formulae' in 'an introduction to R' in your HTML help system. This formula only specifies one term, X, so all you'll get is the result of

y ~ X

If you wanted to fit a polynomial the hard (and not generally good) way you'd have to do 

y ~ I(X )+ I(X^2) + I(X^3) 

Next, in your line
 fit <- lm(Y~conditional.argument)

you have asked lm to fit Y to a character string formed by paste(). That is also unlikely to be useful. You would need to create the complete formula as a string (eg "Y~X+I(X^2)+I(X^3)" ) and then use as.formula to convert that to a formula lm can use.

Then there's the unnecessary loop. You can get your formula string without a loop from an integer 'degree' using 

formstring <- paste("Y ~ ", paste("I(X^",1:degree,")" , sep="", collapse=" + "))

And then you can convert formstring to a formula object and use the formula object in lm:
form <- as.formula(formstring) #be careful with this; it looks for the terms in the current environment... 
lm(form)


Or, if 'degree' were a vector (say 
degree<-c(1, 3, 5)
formstring <- paste("Y ~ ", paste("I(X^",degree,")" , sep="", collapse=" + "))
and so on.

> I am having this problem as well when I want to do a boxplot 
> with variable number of groups to compare.
I don't know what you're trying to achieve with boxplot, but if you gave boxplot itself your vectors it would plot them by itself:

x<-rnorm(17)
y<-runif(23)
z<-c(x,y)

boxplot(x,y,z)

Incidentally, this:
conditional.argument <- data[,1], data[,2], ...,data[,nvariables]          #
is not a valid assignment in R, so it cannot be the code you used. And using something called 'data' (or any other object) in a function that does not have an named argument by the same name is asking for trouble; there is no telling where it will get 'data' from but there is a very good chance it will sometimes be from somewhere you don't expect. 


S Ellison 
 
> -----Original Message-----
> I am sure that my approach is kind of stupid but I cannot 
> think of another way to do it. I apologize...
> 
> foo <- function(..., degree=..){
>   degree<-3 # example
>   
>   conditional.argument <- rep(NA,degree)
>   for(i in 1:(degree-1)) conditional.argument[i] <- 
> paste("X^",i,"+",sep="")
>   conditional.argument[degree] <- paste("X^",degree,sep="")
>   conditional.argument <- paste(conditional.argument,collapse="")
>   
>   conditional.argument
> #  "X^1+X^2+X^3"
> 
>   fit <- lm(Y~conditional.argument)
>   ...
>   return(fit)
> }
> 
> I know that there is the poly() function which I am not looking for.
> I am having this problem as well when I want to do a boxplot 
> with variable number of groups to compare.
> 
> multiple.boxplot <- function(...,nvariables){
> conditional.argument <- data[,1], data[,2], 
> ...,data[,nvariables]          #
> should look like that in the end
> boxplot(conditional.argument)
> }
> 
> Can anyone give me a hint?
> Daniel Hoop
> 
> 
> 
> 
> --
> View this message in context: 
> http://r.789695.n4.nabble.com/Pass-Conditional-loop-argument-i
> nto-a-function-tp4639580.html
> Sent from the R help mailing list archive at Nabble.com.
> 
> ______________________________________________
> 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.
> 

*******************************************************************
This email and any attachments are confidential. Any use...{{dropped:8}}