rowMeans function in dplyr

Question

I have been trying to run the calculate rowMeans within dplyr's mutate function, but keep getting errors. Below is an example DATA set and desired RESULT.

DATA = data.frame(SITE = c("A","A","A","A","B","B","B","C","C"), 
                  DATE = c("1","1","2","2","3","3","3","4","4"), 
                  STUFF = c(1, 2, 30, 40, 100, 200, 300, 5000, 6000),
                  STUFF2 = c(2, 4, 60, 80, 200, 400, 600, 10000, 12000))

RESULT = data.frame(SITE = c("A","A","A","A","B","B","B","C","C"), 
                    DATE = c("1","1","2","2","3","3","3","4","4"), 
                    STUFF = c(1, 2, 30, 40, 100, 200, 300, 5000, 6000),
                    STUFF2 = c(2, 4, 60, 80, 200, 400, 600, 10000, 12000),
                    NAYSA = c(1.5, 3, 45, 60, 150, 300, 450, 7500, 9000))

The code I have written begins by randomly sampling STUFF and STUFF2. Then I would like to calculate the rowMeans of STUFF and STUFF2 and export the result to a new column. I could accomplish this task using tidyr, but would have to redo a larger number of variables. Furthermore I could use the R base package, but prefer to find a solution using the mutate function in dplyr. Thanks in advance.

RESULT = group_by(DATA, SITE, DATE) %>%
  mutate(STUFF=sample(STUFF,replace= TRUE), STUFF2 = sample(STUFF2,replace= TRUE))%>%
  # These approaches return errors 
  mutate(NAYSA = rowMeans(DATA[,-1:-2]))
  mutate(NAYSA = rowMeans(.[,-1:-2])) 
  mutate (NAYSE = rowMeans(.))

Answer 1

You need the rowwise function in dplyr to do that. Your data is random (because of sample) so it produces different results but you will see that it works:

library(dplyr)
  group_by(DATA, SITE, DATE) %>%
  mutate(STUFF=sample(STUFF,replace= TRUE), STUFF2 = sample(STUFF2,replace= TRUE))%>%
  rowwise() %>%
  mutate(NAYSA = mean(c(STUFF,STUFF2)))

Output:

Source: local data frame [9 x 5]
Groups: <by row>

  SITE DATE STUFF STUFF2  NAYSA
1    A    1     1      2    1.5
2    A    1     2      2    2.0
3    A    2    30     80   55.0
4    A    2    30     60   45.0
5    B    3   200    600  400.0
6    B    3   300    200  250.0
7    B    3   100    600  350.0
8    C    4  5000  12000 8500.0
9    C    4  6000  10000 8000.0

As you see it calculates the rowwise mean per row, according to STUFF and STUFF2

Answer 2

@GregF Yep....ungroup() was the key. Thanks.

Working code

RESULT = group_by(DATA, SITE, DATE) %>% 
  mutate(STUFF = sample(STUFF,replace= TRUE), 
         STUFF2 = sample(STUFF2,replace= TRUE)) %>% 
  ungroup() %>% 
  mutate(NAYSA = rowMeans(.[,-1:-2]))