apply statement is not working as expected

[mmarion]

Table.arr consists of a 2x2 matrix of counts followed by a third column of target rates. I would have done the code this way:

Error2.f <- function(mat){
1 - sum(diag(mat[,1:2]))/sum(mat[,1:2]) }
apply(Table.arr,c(1,2),Error2.f) or 
apply(Table.arr,c(1:2),Error2.f)
sd.mat <- apply(Error.arr,2:3,sd)
Mean.mat <- apply(Error.arr,2:3,mean)

I get error - Error in mat[, 1:2] : incorrect number of dimensions

apply(Table.arr,c(3:5),Error2.f)
sd.mat <- apply(Error.arr,2:3,sd)
Mean.mat <- apply(Error.arr,2:3,mean)

Table.arr is randomForest output . It is a list of 2x3 matrices

, , 1, 1, 1, 1

 [,1] [,2]      [,3]

[1,] 152 30 0.1648352
[2,] 55 73 0.4296875

...

I thought the two commas referred to the rows and columns of the output matrices.
Can you help please? Thank you.

MM

[nirgrahamuk]

@mmarion

apply(Table.arr,c(1,2),Error2.f)

your apply says to take each cell in turn MARGINS=c(1,2) and apply a function that expects a matrix input ; but its not a matrix input, its a scalar value, the value of a cell of Table.arr

do you mean to directly apply Error2.f on the 2x2 subset to the top left of Table.arr ?
that would be

Error2.f(Table.arr[1:2,1:2])

[mmarion]

I have just learned about the concept of dimension when working with the apply function. I came up with the following:

Error.arr <- apply(Table.arr,c(3:6),Error2.f)

3:6 refer to dimensions 3,4,5,6. This will change with different analyses.

Your answer is a still another way of looking at the apply statement.
I am getting confused. Can you suggest a source of information about apply?
Youtube videos and white papers are always good.

Thank you for replying.
MM

[nirgrahamuk]

@mmarion

Error.arr <- apply(Table.arr,c(3:6),Error2.f)

Does this work for you or does it give you an error that MARGIN does not match dim(X) ?

[mmarion]

Re: Error.arr <- apply(Table.arr,c(3:6),Error2.f)
This is working.
There is a lot of processing going on and I am studying the dimension issue as I work out the random forest cross validation code. To answer your question and improve on mine will take more effort on my part. I am doing multiple runs of my function and comparing the output. When I have more output I will summarize and may contact RStudio again about this.

I am thinking about your original reply.
Is there a difference in 1:2 and c(1:2) in the apply statement?
Is there a difference in 1,2 and c(1,2) in the apply statement?
Thank you. MM