问题
A few days ago I opened this thread:
Clustering rows by group based on column value
In which we obtained this result:
df <- data.frame(ID = c(1,1,1,1,1,1,1,1,1,1,1, 1, 1,1,1,1,1),
Obs1 = c(1,1,0,1,0,1,1,0,1,0,0,0,1,1,1,1,1),
Control = c(0,3,3,1,12,1,1,1,36,13,1,1,2,24,2,2,48),
ClusterObs1 = c(1, 1, 1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 5))
With:
df <- df %>%
group_by(ID) %>%
mutate_at(vars(Obs1),
funs(ClusterObs1= with(rle(.), rep(cumsum(values == 1), lengths))))
Now I have to make some modifications:
If value of 'Control' is higher than 12 and actual 'Obs1' value is equal to 1 and to previous 'Obs1' value, 'DesiredResultClusterObs1' value should add +1
df <- data.frame(ID = c(1,1,1,1,1,1,1,1,1,1,1, 1, 1,1,1,1,1),
Obs1 = c(1,1,0,1,0,1,1,0,1,0,0,0,1,1,1,1,1),
Control = c(0,3,3,1,12,1,1,1,36,13,1,1,2,24,2,2,48),
ClusterObs1 = c(1, 1, 1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 5),
DesiredResultClusterObs1 = c(1, 1, 1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 5, 6, 6, 6, 7))
I have considered add if_else condition with lag in funs but unsuccessfully, any ideas?
EDIT: How it would be for many columns?
回答1:
This seems to work:
df %>%
mutate(DesiredResultClusterOrbs1 = with(rle(Control > 12 & Obs1 == 1 & lag(Obs1) == 1),
rep(cumsum(values == 1), lengths)) + ClusterObs1)
ID Obs1 Control ClusterObs1 DesiredResultClusterOrbs1
1 1 1 0 1 1
2 1 1 3 1 1
3 1 0 3 1 1
4 1 1 1 2 2
5 1 0 12 2 2
6 1 1 1 3 3
7 1 1 1 3 3
8 1 0 1 3 3
9 1 1 36 4 4
10 1 0 13 4 4
11 1 0 1 4 4
12 1 0 1 4 4
13 1 1 2 5 5
14 1 1 24 5 6
15 1 1 2 5 6
16 1 1 2 5 6
17 1 1 48 5 7
Basically, we use the rle+rep mechanic from your previous thread to create a cumulative vector from the TRUE/FALSE result of your conditions and add it to the existing ClusterObs1.
If you want to create multiple DesiredResultClusterOrbs, you can use mapply. Maybe there's a dplyr solution for this, but this is base R.
Data:
df <- data.frame(ID = c(1,1,1,1,1,1,1,1,1,1,1, 1, 1,1,1,1,1),
Obs1 = c(1,1,0,1,0,1,1,0,1,0,0,0,1,1,1,1,1),
Obs2 = rbinom(17, 1, .5),
Control = c(0,3,3,1,12,1,1,1,36,13,1,1,2,24,2,2,48),
ClusterObs1 = c(1, 1, 1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 5))
df <- df %>%
mutate_at(vars(Obs2),
funs(ClusterObs2= with(rle(.), rep(cumsum(values == 1), lengths))))
The loop:
newcols <- mapply(function(x, y){
with(rle(df$Control > 12 & x == 1 & lag(x) == 1),
rep(cumsum(values == 1), lengths)) + y
}, df[2:3], df[5:6])
This produces a matrix with the new columns, which you can then rename and cbind to your data:
colnames(newcols) <- paste0("DesiredResultClusterOrbs", 1:2)
cbind.data.frame(df, newcols)
ID Obs1 Obs2 Control ClusterObs1 ClusterObs2 DesiredResultClusterOrbs1 DesiredResultClusterOrbs2
1 1 1 1 0 1 1 1 1
2 1 1 1 3 1 1 1 1
3 1 0 0 3 1 1 1 1
4 1 1 0 1 2 1 2 1
5 1 0 0 12 2 1 2 1
6 1 1 0 1 3 1 3 1
7 1 1 1 1 3 2 3 2
8 1 0 0 1 3 2 3 2
9 1 1 1 36 4 3 4 3
10 1 0 1 13 4 3 4 4
11 1 0 0 1 4 3 4 4
12 1 0 1 1 4 4 4 5
13 1 1 1 2 5 4 5 5
14 1 1 0 24 5 4 6 5
15 1 1 1 2 5 5 6 6
16 1 1 1 2 5 5 6 6
17 1 1 1 48 5 5 7 7
来源：https://stackoverflow.com/questions/51038794/clustering-rows-by-group-based-on-column-value-with-conditions