[R] dplyr instead of plyr: Help

[Axel Urbiz]

 Sorry, this is *related* to a recent post, but not the same. I'd
appreciate your help in getting the same results with the two methods below
(the first using plyr and the second using dplyr. The former works, but not
the latter.)

### Sample data

set.seed(4)

df <- data.frame(pred = rnorm(100), y = sample(c(0,1), 100, replace = TRUE),

                 models = gl(2, 50, 100, labels = c("model1", "model2")))

### using plyr


fooFun <- function(x) {

  xcuts <- unique(x$pred)

  x$bin <- cut(x$pred, breaks = xcuts, include.lowest = TRUE)

  x <- dplyr::summarize(dplyr::group_by(x, bin), sumY = sum(y))

  x

}

head(plyr::ddply(df, plyr::.(models), fooFun))


### Using dplyr


fooFun2 <- function(pred, y) {

  xcuts <- unique(pred)

  bin <- cut(pred, breaks = xcuts, include.lowest = TRUE)

  dft <- data.frame(bin, pred, y)

  dft <- dplyr::summarize(dplyr::group_by(dft, bin), sumY = sum(y))

  dft

}

res_dplyr <- dplyr::mutate(dplyr::group_by(df, models), fooFun2(pred, y))

head(res_dplyr)


Thanks

Axel.

	[[alternative HTML version deleted]]