Last observation carried forward conditional on value and colums - r

Question

I have a longhitudinal dataframe with a lot of missing values that looks like this.
ID = c(1,1,1,1,1,2,2,2,2,2,3,3,3,3,3)
date = c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5)
cond = c(0,0,0,1,0,0,0,0,1,0,0,0,0,0,0)
var = c(1, NA , 2, 0,NA, NA, 3, NA,0, NA, 2, NA, 1,NA,NA)
df = data.frame(ID, date, cond,var)
I would like to carry forward the last observation based on two conditions:
1) when cond=0 it should carry on the observation the higher value of the variable of interest.
2) when cond=1 it should carry forward the lower value of the variable of interest.
Does anyone have an idea on how I could do this in an elegant way?
The final dataset should look like this
ID = c(1,1,1,1,1,2,2,2,2,2,3,3,3,3,3)
date = c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5)
cond = c(0,0,0,1,0,0,0,0,1,0,0,0,0,0,0)
var = c(1, 1 , 2, 0, 0, NA, 3, 3, 0, 0,2,2,2,2,2)
final = data.frame(ID, date, cond,var)
So far I was able to carry forward the last observation, but I was unable to impose the conditions
library(zoo)
df <- df %>%
group_by(ID) %>%
mutate(var =
na.locf(var, na.rm = F))
any suggestion is welcomed

Answer 1

This is the use of accumulate2 ie
df%>%
group_by(ID)%>%
mutate(d = unlist(accumulate2(var,cond[-1],function(z,x,y) if(y) min(z,x,na.rm=TRUE) else max(z,x,na.rm=TRUE))))
# A tibble: 15 x 5
# Groups: ID [3]
ID date cond var d
<dbl> <dbl> <dbl> <dbl> <dbl>
1 1 1 0 1 1
2 1 2 0 NA 1
3 1 3 0 2 2
4 1 4 1 0 0
5 1 5 0 NA 0
6 2 1 0 NA NA
7 2 2 0 3 3
8 2 3 0 NA 3
9 2 4 1 0 0
10 2 5 0 NA 0
11 3 1 0 2 2
12 3 2 0 NA 2
13 3 3 0 1 2
14 3 4 0 NA 2
15 3 5 0 NA 2

Answer 2

I think, if I understand what you are after is this?
ID = c(1,1,1,1,1,2,2,2,2,2,3,3,3,3,3)
date = c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5)
cond = c(0,0,0,1,0,0,0,0,1,0,0,0,0,0,0)
var = c(1, NA , 2, 0,NA, NA, 3, NA,0, NA, 2, NA, 1,NA,NA)
df = data.frame(ID, date, cond,var)
Using case_when you can do some conditional checks. I'm unsure if you mean to return the minimum for all of the "ID" field, but this will look at the condition and then lag or lead to find a non missing value
library(dplyr)
df %>%
mutate(var_imput = case_when(
cond == 0 & is.na(var)~lag(x = var, n = 1, default = NA),
cond == 1 & is.na(var)~lead(x = var, n = 1, default = NA),
TRUE~var
))
Which yields:
ID date cond var var_imput
1 1 1 0 1 1
2 1 2 0 NA 1
3 1 3 0 2 2
4 1 4 1 0 0
5 1 5 0 NA 0
6 2 1 0 NA NA
7 2 2 0 3 3
8 2 3 0 NA 3
9 2 4 1 0 0
10 2 5 0 NA 0
11 3 1 0 2 2
12 3 2 0 NA 2
13 3 3 0 1 1
14 3 4 0 NA 1
15 3 5 0 NA NA
If you want to group by ID then you could generate an impute table by ID, then join it with the original table like this:
# enerate input table
input_table <- df %>%
group_by(ID) %>%
summarise(min = min(var, na.rm = T),
max = max(var, na.rm = T)) %>%
gather(cond, value, -ID) %>%
mutate(cond = ifelse(cond == "min", 0, 1))
# Join and impute missing
df %>%
left_join(input_table,by = c("ID", "cond")) %>%
mutate(var_imput = ifelse(is.na(var), value, var))