Mode in R by groups

Question

I need to calculate the mode of an identity number for each group of ages. Let\'s suposse the following table:

library(data.table)
DT = data.table(age=c(12,1         


        

Answer 1

One approach:

> myfun <- function(x) unique(x)[which.max(table(x))]
> DT[ , moda := myfun(number), by = age]
> DT
   age          v number moda
1:  12 -0.9740026    122  122
2:  12  0.6893727    125  122
3:   3 -0.9558391      5    5
4:   3 -1.2317071      5    5
5:  12 -0.9568919    122  122

Answer 2

You can use dplyrfor this:

library(dplyr)
modes_by_age <- summarise(group_by(DT, age), group_mode = g(number))
inner_join(DT, modes_by_age)

This gives your desired output:

Source: local data table [5 x 4]

  age         v number group_mode
1   3 0.5524352      5          5
2   3 0.2869912      5          5
3  12 0.8987475    122        122
4  12 0.9740715    125        122
5  12 2.5058450    122        122

Answer 3

modef <- function(V)
{ 
k = 1
prev='xxxx'
max_value = 0
for (i in V)
{
if (prev == i)
{ 
k = k+1
}
else
{
if (k > max_value) 
{
MODE_CALC = data.frame( 
number = c(prev) , 
occurence = c(k) )
max_value = k
k = 1
}
k = 1
}
prev = i
}
print(MODE_CALC$number)
}
V = c(11, 11, 11, 11, 12, 12, 2, 2, 2, 2, 2, 2, 14, 14, 14, 15, 16, 17, 17, 17 ,17 , 
17, 18, 19)    
modef(sort(V))

Answer 4

Here's a base R solution. You could compute the mode for each group and then merge with your original data:

merge(DT, setNames(aggregate(number~age, data=DT, g), c("age", "moda")), by="age")
#    age          v number moda
# 1:   3  1.7148357      5    5
# 2:   3  0.9504811      5    5
# 3:  12 -0.7648237    122  122
# 4:  12  0.9011115    125  122
# 5:  12 -0.8718779    122  122

There may be a data table-specific approach, but this would work even if DT were a data frame.