在R函数中使用for循环

Question

Goal: take a data.frame with headers and return a new data.frame with additional variables created from calculations within a function.

My existing code works for creating a transform of a data.frame:

transform<- function(x) {
  transformtemp<- x
  transformtemp[1]<- x[1]
  for(i in 2:length(x)) {
    transformtemp[i]<- x[i] + 0.9*transformtemp[i-1]
  }
  transformscale<- sum(x)/sum(transformtemp)
  x<- transfomrtemp*transformscale
}

x[]<- lapply(x,transform)

With this code, I get back a data.frame with the function applied to all columns of my data.

I need Help with: 1. As of now, this code only uses 0.9 as my decay parameter. I want to create output using more decay parameters, say decay<- seq(0,1,0.1) and save them for use. 2. I want the output to be the original data plus new columns of data with the function applied at the different decay rates, with names like "column1_0.9", "column1_0.8", "column2_0.9" etc.

I have tried using another loop with a changing decay rate but can't seem to get it right. I hope this all makes sense and let me know if I need to clarify further.

All the best and thanks!

Answer 1

You function without assigning to the original data:

mytransform<- function(x) {
  transformtemp<- x
  transformtemp[1]<- x[1]
  for(i in 2:length(x)) {
    transformtemp[i]<- x[i] + 0.9*transformtemp[i-1]
  }
  transformscale<- sum(x)/sum(transformtemp)
  x<- transformtemp*transformscale
}

A <- do.call(data.frame, lapply(mtcars,mytransform))

Let's make it more efficient with the filter function:

mytransform1 <- function(x, d) {
  y <- filter(x, d, sides=1, method="recursive")

  transformscale<- sum(x)/sum(y)
  c(y*transformscale)
}

B <- do.call(data.frame, lapply(mtcars, mytransform1, d=0.9))

all.equal(A, B)
#[1] TRUE

Now we add support for multiple decay rates with an additional sapply loop:

mytransform <- function(x, d) {
  y <- sapply(d, function(d, x) c(filter(x, d, sides=1, method="recursive")), x=x)
  transformscale<- sum(x)/colSums(y)
  t(t(y)*transformscale)
}
C <- do.call(data.frame, lapply(mtcars, mytransform, d=c(0.9, 0.8)))
all.equal(B, setNames(C[,seq_along(B)*2-1], names(B)))
#[1] TRUE