CODEWITHSUNDEEP
rdplyrgrouping
Curtis
Curtis

Reputation: 181

Get Max Value per Run or Series in Sequence

I am trying to get a a max value per stretch of an indicator, or repeating value.

Here is an example:

A = c(28, 20, 23, 30, 26, 23, 25, 26, 27, 25, 30, 26, 25, 22, 24, 25, 24, 27, 29)
B = c(0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 0, 0, 0, 1)


df <- as.data.frame(cbind(A, B))
df
A B
28 0
20 1
23 1
30 0
26 0
23 1
25 1
26 1
27 0
25 0
30 1
26 1
25 1
22 0
24 1
25 0
24 0
27 0
29 1

For each group or stretch of 1's in column B I want to find the max in column A. The max column could be an indicator that A it is a max or the actual value in A, and be NA or 0 for other values of B.

The output I am hoping for looks something like this:

A B  max
28 0 0
20 1 0
23 1 1
30 0 0
26 0 0
23 1 0
25 1 0
26 1 1
27 0 0
25 0 0
30 1 1
26 1 0
25 1 0
22 0 0
24 1 1
25 0 0
24 0 0
27 0 0
29 1 1

I've tried to generate groups per section of column B that = 1 but I did not get very far because most grouping functions require unique values between groups.

Also, please let me know if there are any improvements to the title for this problem.

Upvotes: 1

Views: 159

Answers (2)

akrun
akrun

Reputation: 887088

One option would be data.table

library(data.table)
setDT(df)[, Max := +((A== max(A)) & B), rleid(B) ]
df
#     A B Max
# 1: 28 0   0
# 2: 20 1   0
# 3: 23 1   1
# 4: 30 0   0
# 5: 26 0   0
# 6: 23 1   0
# 7: 25 1   0
# 8: 26 1   1
# 9: 27 0   0
#10: 25 0   0
#11: 30 1   1
#12: 26 1   0
#13: 25 1   0
#14: 22 0   0
#15: 24 1   1
#16: 25 0   0
#17: 24 0   0
#18: 27 0   0
#19: 29 1   1

Or as @Frank mentioned, for better efficiency, we can make use gmax by first assigning column and then replace

DT[, MA := max(A), by=rleid(B)][A == MA & B, Max := 1L][]

Upvotes: 3

pogibas
pogibas

Reputation: 28339

Solution using dplyr

library(dplyr)
df %>% 
    group_by(with(rle(B), rep(seq_along(lengths), lengths))) %>% 
    mutate(MAX = ifelse(B == 0, 0, as.numeric(A == max(A)))) %>% 
    .[, c(1, 2, 4)]

       A     B   MAX
   <dbl> <dbl> <dbl>
 1    28     0     0
 2    20     1     0
 3    23     1     1
 4    30     0     0
 5    26     0     0
 6    23     1     0
 7    25     1     0
 8    26     1     1
 9    27     0     0
10    25     0     0
11    30     1     1
12    26     1     0
13    25     1     0
14    22     0     0
15    24     1     1
16    25     0     0
17    24     0     0
18    27     0     0
19    29     1     1

Upvotes: 1

Related Questions