Comparing columns from different rows in R - r

Question

In R using dplyr I'm struggling to accumulate two columns through a sequence.
What I'd like to do:
Within each Outlet I'm trying to calculate the cumulative DFLSEcr (cumulative DFLSEcr = cumu_DFLSEcr) and count (cumu_count) for each row based on a sequence of ZHYD and NextDown. Each row has a value for NextDown which corresponds to the row which comes before it signified by a matching ZHYD. This makes a sequence in which DFLSEcr and count accumulate. Where Exutoire == 0 then cumu_DFLSEcr == 0 and cumu_count == 0. If DFLSEcr == 1 or NA then don't include it in the sum. I've used lag() but I don't think this is correct...
Input:
input <- structure(list(ZHYD = structure(c(1L, 2L, 3L, 4L, 5L, 6L, 7L,
10L, 8L, 9L, 11L), .Label = c("B020006183", "B020006184", "B020006185",
"B020006190", "B020006199", "B020006212", "B020006228", "B020006278",
"B020006285", "B020006290", "B020006325"), class = "factor"),
Outlet = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L,
2L, 3L), .Label = c("BSO0001727", "BSO0001746", "BSO0001756"
), class = "factor"), NextDown = structure(c(1L, 1L, 2L,
2L, 3L, 3L, NA, NA, 4L, 4L, 5L), .Label = c("B020006190",
"B020006199", "B020006228", "B020006290", "B020006335"), class = "factor"),
count = c(15L, 55L, 42L, 19L, 32L, 6L, 19L, 49L, 4L, 82L,
5L), DFLSEcr = c(0.07, 0.02, 0.02, 0.05, 0.03, 0.17, 0.05,
0.02, 0.25, 0.01, NA), Exutoire = c(0L, 0L, 0L, 0L, 0L, 0L,
1L, 1L, 0L, 0L, 0L)), class = "data.frame", row.names = c(NA,
-11L), .Names = c("ZHYD", "Outlet", "NextDown", "count", "DFLSEcr",
"Exutoire"))
Method tried so far:
input %>%
group_by(Outlet) %>%
select(ZHYD, NextDown, ZHYD, Outlet, Exutoire, count, DFLSEcr) %>%
mutate(cleanDFLSEcr = replace(DFLSEcr, DFLSEcr == 1, 0),
cleanDFLSEcr = replace(DFLSEcr, is.na(DFLSEcr), 0),
cumu_DFLSEcr = if_else(Exutoire == 1, 0, cumsum(cleanDFLSEcr)) %>% lag(1, default = 0),
cumu_count = if_else(Exutoire == 1, 0, as.numeric(cumsum(count))) %>% lag(1, default = 0)) %>%
select(-cleanDFLSEcr)
Desired output:
ZHYD Outlet NextDown count DFLSEcr Exutoire cumu_count cumu_DFLSEcr
<fct> <fct> <fct> <int> <dbl> <int> <int> <dbl>
1 B020006183 BSO0001727 B020006190 15 0.0700 0 70 0.130
2 B020006184 BSO0001727 B020006190 55 0.0200 0 70 0.130
3 B020006185 BSO0001727 B020006199 42 0.0200 0 51 0.0800
4 B020006190 BSO0001727 B020006199 19 0.0500 0 51 0.0800
5 B020006199 BSO0001727 B020006228 32 0.0300 0 19 0.0500
6 B020006212 BSO0001727 B020006228 6 0.170 0 19 0.0500
7 B020006228 BSO0001727 <NA> 19 0.0500 1 0 0.
8 B020006290 BSO0001746 <NA> 49 0.0200 1 0 0.
9 B020006278 BSO0001746 B020006290 4 0.250 0 1 0.0200
10 B020006285 BSO0001746 B020006290 82 0.0100 0 1 0.0200
11 B020006325 BSO0001756 B020006335 5 NA 0 1 0.0200
cumu_count and cumu_DFLSEcr outputs are the same in some cases because they share the same NextDown.
The final row in the sequence shouldn't be included in cumu_count. So for the first row cumu_count == 19 +32 + 19 = 70
Edit
Turns out I needed igraph as this is a routing problem.

Answer 1

Ok your problem is not simple. You have nested data which is a bit complicated to manipulate.
I give you an answer. There are for sure better ways to do it. But it can give you some ideas.
library(dplyr)
library(tidyr)
father_son_table <- select(input, actual = ZHYD, father = NextDown)
sequences <- rename(input, actual = ZHYD, father = NextDown) %>%
left_join(father_son_table, by = c("father" = "actual"), suffix= c(".1", ".2")) %>%
left_join(father_son_table, by = c("father.2" = "actual"), suffix = c(".1", ".3")) %>%
tibble::rowid_to_column(var = "sequence_number")
table_order <- sequences %>%
select(-count, -DFLSEcr, -Exutoire, -Outlet) %>%
gather(key = height, value = node, -sequence_number) %>%
mutate(order = case_when( height == "actual" ~ 0,
height =="father.1" ~ 1,
height == "father.2" ~ 2,
height == "father.3" ~ 3 )) %>%
na.omit() %>%
select(sequence_number, node, order)
result <- left_join(table_order, input, by = c("node" = "ZHYD")) %>%
arrange(sequence_number, order) %>%
group_by(sequence_number, Outlet) %>%
mutate(cumu_count = sum(count) - count,
cumu_DFLSE_cr = sum(DFLSEcr)- DFLSEcr) %>%
filter(order == 0)