Create class column in data.table given a numeric vector

Question

Currently I have a data.table like this:

item    dummyvar  
4q7C0o     1         
2BrKY63    1         
3drUy6I    1         
G5ALtO    1000        
G5G859    1000
PAP589    2000

And using a defined function I find out that the rows where there is a significant change in the dummyvar are in a numeric vector called imbalance = 4 6. What I would like to do is to create a new column in my data table such that all the rows below the given numbers given by imbalance are in a given class, for ex. something like this:

item    dummyvar   Class
4q7C0o     1         1
2BrKY63    1         1
3drUy6I    1         1
G5ALtO    1000       2 
G5G859    1000       2
PAP589    2000       3

Okay, I gave an answer an realised that I did not understand your `imbalance`-part. Could you be more precise? I just numerated the groups, but don't know what you are trying to do. — Martin Gal, Jun 01 '20 at 15:47
Perhaps `your_data_table[, Class := cumsum(.I %in% imbalance)]`? I'm assuming `imbalance` is a vector of the row numbers where you want `Class` to increment. (This version starts at 0, but you could stick a `+ 1` on it to start at `1`) — Gregor Thomas, Jun 01 '20 at 15:51
Related / possible duplicate: [*How to create a consecutive index based on a grouping variable in a dataframe*](https://stackoverflow.com/q/6112803/2204410) — Jaap, Jun 01 '20 at 16:00
@MartinGal, `imbalance` is a numeric vector where each number tells the row where the class should change, for ex. rows 1-3 should be class 1, and then rows 4-5 will be class 2 — Oliver, Jun 01 '20 at 16:00
@MartinGal on the contrary, thank you for your time for commenting and checking my problem. I appreciate it — Oliver, Jun 01 '20 at 16:08
@GregorThomas Thank you, That is what I was looking for. I appreciate your help — Oliver, Jun 01 '20 at 16:09

Uwe · Answer 1 · 2020-06-01T16:20:32.817

If I understand correctly, the vector imbalance contains the row indices into the data.table at which a new Class should start.

Here is one possible solution (among many other, I suppose) using cut() on the row index .I:

dt[, Class := cut(.I, c(0, imbalance, Inf), labels = FALSE, right = FALSE)][]

      item dummyvar Class
1:  4q7C0o        1     1
2: 2BrKY63        1     1
3: 3drUy6I        1     1
4:  G5ALtO     1000     2
5:  G5G859     1000     2
6:  PAP589     2000     3

score 1 · Accepted Answer · answered Jun 01 '20 at 16:28

1

Perhaps your_data_table[, Class := cumsum(.I %in% imbalance)]? I'm assuming imbalance is a vector of the row numbers where you want Class to increment. (This version starts at 0, but you could stick a + 1 on it to start at 1)

answered Jun 01 '20 at 16:28

Gregor Thomas

136,190
20
167
294

Create class column in data.table given a numeric vector

2 Answers2