CODEWITHSUNDEEP
rdata.tableminimum
DJack
DJack

Reputation: 4940

How can you get all the minimum values of a variable by group?

I have a data frame:

df<-data.frame(P = c("A","A","A", "B","B","B", "C", "C", "C"), 
               index = c("ind1","ind2","ind3","ind1","ind2","ind3","ind1","ind2","ind3"),
               var = c(2,1,1,8,5,4,2,8,6))

I would like to get ALL the minimum valueS of var and their associated index for each values of P. I can do this:

DT <- data.table(df)
DT[  ,.SD[which.min(var)], by = P]

which gives only one minimum value of var (the first one) by P:

   P index  var
1: A  ind2   1
2: B  ind3   4
3: C  ind1   2

And I would like:

   P index  var
1: A  ind2   1
2: A  ind3   1
2: B  ind3   4
3: C  ind1   2

Ideas?

Upvotes: 3

Views: 152

Answers (2)

talat
talat

Reputation: 70336

Using dplyr, you could use one of the following:

library(dplyr)
DT %>% group_by(P) %>% filter(var == min(var))  # or %in% instead of ==
#Source: local data table [4 x 3]
#Groups: P
#
#       P  index   var
#  (fctr) (fctr) (dbl)
#1      A   ind2     1
#2      A   ind3     1
#3      B   ind3     4
#4      C   ind1     2

Or

DT %>% group_by(P) %>% top_n(1, desc(var)) # top_n() returns multiple rows in case of ties
#Source: local data table [4 x 3]
#Groups: P
#
#       P  index   var
#  (fctr) (fctr) (dbl)
#1      A   ind2     1
#2      A   ind3     1
#3      B   ind3     4
#4      C   ind1     2

Or

DT %>% group_by(P) %>% filter(min_rank(var) == 1)
#Source: local data table [4 x 3]
#Groups: P
#
#       P  index   var
#  (fctr) (fctr) (dbl)
#1      A   ind2     1
#2      A   ind3     1
#3      B   ind3     4
#4      C   ind1     2

Upvotes: 5

A5C1D2H2I1M1N2O1R2T1
A5C1D2H2I1M1N2O1R2T1

Reputation: 193687

From the help page for which.min, you'll note that it says:

Determines the location, i.e., index of the (first) minimum or maximum of a numeric (or logical) vector.

If you wanted all values that match the minimum, you should try using ==. Thus, continuing with your approach, try:

DT[, .SD[var == min(var)], by = P]
##    P index var
## 1: A  ind2   1
## 2: A  ind3   1
## 3: B  ind3   4
## 4: C  ind1   2

Upvotes: 4

Related Questions