Problems working with factors and apply functions

Question

What I have is a data frame that contains, among others, a factor field which holds a range of values used as factor. From what I understand it is essentially bins for numeric values.

What I want to do is to convert these to numeric values so I can use them in the downstream analysis. The idea is simple enough; (a) get a function that takes the factor level, split it at the dash and extract numeric values and calculates the average and (b) apply the function of the column

data$Range.mean <- sapply(data$Range, function(d) { range <- as.matrix(strsplit(as.character(d), "-")) (as.numeric(range[,1]) + as.numeric(range[,2]))/2 })

Which gives the following error

Error in FUN(X[[1L]], ...) : (list) object cannot be coerced to type 'double'

I tried lapply instead which makes no difference. While looking for answers, I found some other solutions to this problem, which is essentially extracting the lower and upper bound separately to individual arrays then of course calculating pairwise average is trivial.

I would like to understand what I am doing/thinking wrong here though. Why is my code giving an error, and what does that error mean, really?

@alexis_laz unlist seems to do the trick, together with changing the way I access the first and second element. Although I am still a bit confused as to why I got the error message in the first place. The error message is a bit cryptic to me. — posdef
– posdef, Commented Nov 12, 2013 at 12:45
In ?strsplit you'll see that the value of this function is a list. as.numeric tries to be applied on a "list-y" object and gives this error. — alexis_laz
– alexis_laz, Commented Nov 12, 2013 at 12:48

Backlin · Accepted Answer · 2013-11-12 12:56:44Z

You are correct in that factors in fact are integers with labeled bins. So if you have a factor like this

x <- factor(c("0-1", "0-1", "1-2", "1-2"))

it is essentially a combination of the following components

as.integer(x) levels(x)

To convert the factor to the actual values specified by its lables, you can take a detour through as.character and parse that into numbers.

# Recreating a data frame with a factor like yours data <- data.frame(Range = cut(runif(100), 0:10/10)) levels(data$Range) <- sub("\\((.*),(.*)]", "\\1-\\2", levels(data$Range)) # Calculating range means sapply(strsplit(as.character(data$Range), "-"), function(x) mean(as.numeric(x)))

Collectives™ on Stack Overflow

Problems working with factors and apply functions

1 Answer 1

Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Related