CODEWITHSUNDEEP
rfunction
Arrebimbomalho
Arrebimbomalho

Reputation: 186

Modifying a function to return an expression based on the values of a column

Suppose I have the following Dataframe (reproducible example follows):

DF <- data.frame(ID=c(12345,12345,13425),NumTrans=c("1-2","2-3","1-3"),Cov1=c(1500,1000,2000),cov2=c(1,0,1))

The field NumTrans represents transitions between states: subject 12345, say, transitions between states 1 and 2, and so on. Writing the function "func" as follows, 

func=function(x){
a=paste0("exp(",x[1],"*x","+",x[2],"*y)")
return(a)
}

allows you to associate to each ID an expression outputting exp(Cov1*x1 + Cov2*y):

by(DF[3:4],DF[1],func)

ID: 12345
[1] "exp(c(1500, 1000)*x+c(1, 0)*y)"
-----------------------------------------
ID: 13425
[1] "exp(2000*x+1*y)"

However, what I need is slightly different.

1) Is there any way to modify this function such that on the "*x" and "*y" you obtain instead "*x_NumTrans" and "*y_NumTrans", where NumTrans is the NumTrans of that row?

2) Is it possible to write a function that returns this output, but equal to the number of rows?

In other words, the desired output is something like:

[1] "exp(c(1500, 1000)*x_1-2+c(1, 0)*y_1-2)"
----------------------------------------------
[2] "exp(c(1500, 1000)*x_2-3+c(1, 0)*y_2-3)"
----------------------------------------------
[3] "exp(2000*x_1-3+1*y_1-3)"
----------------------------------------------

Or even better, a solution that puts each of these expressions in a column that can be appended to DF...

Many thanks in advance!

Upvotes: 0

Views: 42

Answers (1)

chinsoon12
chinsoon12

Reputation: 25225

You can do it as follows. Split into 2 sections. The first for those with c(*,*) and then paste the results with NumTrans

byAns <- by(DF[2:4], DF[1], function(x) {
    s1 <- paste0("exp(", x[2], "*x_")
    s2 <- paste0("+", x[3], "*y_")
    paste0(s1, x[[1]], s2, x[[1]], ")")
})

Upvotes: 1

Related Questions