使用 data.table 在关系计数中不等于

Question

I am hoping to understand how to use data.table to calculate a count given all levels of a particular categorical variable that do not match the value for the record.

Take the follow data.table.

df <- data.table(var1 = c('dog','cat','dog','cat','dog','dog','dog'),
             var2 = c(1,5,90,95,91,110,8),
             var3 = c('lamp','lamp','lamp','table','table','table','table'))

I would like to calculate the count of values that fall within a range and do not share the same value for var1 .

This is related to Count of values within specified range of value in each row using data.table . To quote the answer from @Jaap, the following code allows for producing a count within a range.

df[, var2withinrange := df[.(var2min = var2 - 5, var2plus = var2 + 5)
                       , on = .(var2 >= var2min, var2 <= var2plus)
                       , .N
                       , by = .EACHI][, N]][]

In attempting to expand this answer, I had success in requiring an exact match for var1 with the following:

df[, var2withinrange := df[.(var2min = var2 - 5, var2plus = var2 + 5, var1 = var1)
                       , on = .(var2 >= var2min, var2 <= var2plus, var1 = var1)
                       , .N
                       , by = .EACHI][, N]][]

The code below is my attempt at counting when var1 holds a value that is not equal to the var1 value in the given row, but this code fails.

df[, var2withinrange := df[.(var2min = var2 - 5, var2plus = var2 + 5, var1 = var1)
                       , on = .(var2 >= var2min, var2 <= var2plus, var1 != var1)
                       , .N
                       , by = .EACHI][, N]][]

How can a "not equal to" type operator be added? A data.table answer is preferable, but of course a solution in dplyr or really any alternative would be appreciated!

Answer 1

In this particular case, you can do the following:

df[.(var2min = var2 - 5, var2plus = var2 + 5, v1=var1)
    , on = .(var2 >= var2min, var2 <= var2plus)
    , sum(v1 != x.var1)
    , by = .EACHI]

output:

   var2 var2 V1
1:   -4    6  1
2:    0   10  2
3:   85   95  1
4:   90  100  2
5:   86   96  1
6:  105  115  0
7:    3   13  1

In general, I think you can do an anti-join.