CODEWITHSUNDEEP
rdataframe
Vahid Mirjalili
Vahid Mirjalili

Reputation: 6501

R extract the n-th lowest value by values of one column, for all the levels of another column

I have this data frame:

> df
   c1 c2
1   1  b
2   2  a
3   3  a
4   4  a
5   3  a
6   2  b
7   6  a
8   4  b
9   8  b
10  7  a

So for i=3, the output should be the 3rd lowest values of "c1" for both levels of "c2"

   c1  c2
   3   a
   4   b

Upvotes: 3

Views: 167

Answers (3)

Michele
Michele

Reputation: 8753

You have different options, one can be tapply

    > df<-read.table(text="   c1 c2
 1   1  b
 2   2  a
 3   3  a
 4   4  a
 5   3  a
 6   2  b
 7   6  a
 8   4  b
 9   8  b
 10  7  a")
> df
   c1 c2
1   1  b
2   2  a
3   3  a
4   4  a
5   3  a
6   2  b
7   6  a
8   4  b
9   8  b
10  7  a
> tapply(df$c1, df$c2, function(x) sort(x)[3])
a b 
3 4

Or, using plyr package, you can:

> library(plyr)

> ddply(df, .(c2), summarise, c1=sort(c1)[3])
  c2 c1
1  a  3
2  b  4

Upvotes: 3

Metrics
Metrics

Reputation: 15458

Using data.table

library(data.table)
dt<-data.table(df1)
dt[,sort(c1)[3],by=c2]
   c2 V1
1:  b  4
2:  a  3

Upvotes: 1

A5C1D2H2I1M1N2O1R2T1
A5C1D2H2I1M1N2O1R2T1

Reputation: 193507

Essentially a variation on a theme:

aggregate(c1 ~ c2, df, function(x) sort(x)[3])
#   c2 c1
# 1  a  3
# 2  b  4

Upvotes: 4

Related Questions