2

I have a data table where I want to standardize the name in "Sex" and calculate the geometric mean based on each Group (as in x, y and z in the table).

Would appreciate your help. Below are the data.table.

library(data.table)
dt <- data.table(Group = c("x","x","x","y","z","z"), Sex = c("Man","Female","Feminine","Male","M","F"), Score = c(0,0.4,0.1,0.5,3,2.1))

Thank you.

Twinstan
  • 23
  • 3
  • Please share what you have already tried, rather than asking for someone to write your code for you – Zac Apr 20 '20 at 14:23

1 Answers1

1

Is this what you want?

geomean <- function(v) prod(v)**(1/length(v))
res <- tapply(dt$Score, dt$Group, geomean)

which gives

> res
      x       y       z 
0.00000 0.50000 2.50998

or use ave to create a new column

dt <- within(dt,gm <- ave(Score,Group,FUN = geomean))
> dt
Group      Sex Score      gm
1:     x      Man   0.0 0.00000
2:     x   Female   0.4 0.00000
3:     x Feminine   0.1 0.00000
4:     y     Male   0.5 0.50000
5:     z        M   3.0 2.50998
6:     z        F   2.1 2.50998

EDIT:

If you want to group data by both Group and Sex, try below

dt <- within(transform(dt,Sex = toupper(substr(Sex,1,1))),
             gm <- ave(Score,Group,Sex,FUN = geomean))

thus

> dt
   Group Sex Score  gm
1:     x   M   0.0 0.0
2:     x   F   0.4 0.2
3:     x   F   0.1 0.2
4:     y   M   0.5 0.5
5:     z   M   3.0 3.0
6:     z   F   2.1 2.1
ThomasIsCoding
  • 96,636
  • 9
  • 24
  • 81