CODEWITHSUNDEEP
r
QBarber
QBarber

Reputation: 25

How to use lapply to find closest value in a list in R?

I'm trying to find the model-predicted value closest to a real observed value within a large dataframe. I believe I need to use lapply, but I'm really not sure. Thanks in advance, SE, and sorry if this is a repeat of a previous post, I looked.

df <- data.frame(pred = rnorm(50, mean = 100, sd = 10),
                 cand = I(replicate(50, exp = I(list(rnorm(6, mean = 100, sd = 10))))))

So far, I've come up with a 1-line function that works when run on a single row, but I have two problems:

df$closest <- sapply( df, function(x) { which.min( abs( df$pred[x] - df$cand[[x]] ) ) } )

  1. This function won't work on the full list, probably because I am new to the apply family.

  2. This function returns a list position, not the actual value, which is what I need.

    Error in df$cand[[x]] : no such index at level 1

Upvotes: 2

Views: 179

Answers (2)

Badger
Badger

Reputation: 1053

apply allows us to operate on the rows, or the columns, because you are looking to loop through the rows, a margin of 1 (rows) should get the job done!

We could use apply: 

df$closest <- apply( df,MARGIN = 1, function(x) { which.min( abs( x$pred - x$cand ) ) } )

Upvotes: 1

akrun
akrun

Reputation: 887128

Here, we can use Map instead of sapply because sapply loops over each of the columns and the x anonymous function value is the value of that column. It cannot be used for indexing

df$closest <-  unlist(Map(function(x, y) which.min(abs(y - x)), df$pred, df$cand))

Or else with sapply, we have to loop over the row index

Upvotes: 0

Related Questions