How to select the highest absolut value in a column?-CodePudding

In the dataframe

Name <- c("Jon", "Jon", "Maria", "Maria","dan","dan","Jon", "Jon", "Maria", "Maria","dan","dan")
tem <- c(-23, -41, 32, 58,30,10,-20,63,-30,-33,55,66)
group <- c("max", "min", "max", "min","max","min","max", "min", "max", "min","max","min")
var <- c("lan", "lan", "lan", "lan","lan","lan","sea", "sea", "sea", "sea","sea","sea")
df <- data.frame(Name, tem,group,var)

I need to find the highest absolute tem by name and by group

for instance

for Jon max, lan is higher than sea, so I keep lan.

for Jon min sea is higher than lan, so I keep sea

desired output

 Jon -23   max lan
Maria  32   max lan
Maria  58   min lan
Jon  63   min sea
dan  55   max sea
dan  66   min sea

CodePudding user response：

Looks about right

do.call(
  rbind,
  by(df,list(df$Name,df$group),function(x){
    x[which.max(abs(x$tem)),]
  })
)

    Name tem group var
11   dan  55   max sea
1    Jon -23   max lan
3  Maria  32   max lan
12   dan  66   min sea
8    Jon  63   min sea
4  Maria  58   min lan

CodePudding user response：

You may use which.max with abs to get index of maximum value.

library(dplyr)

get_abs_max <- function(x) which.max(abs(x))

df %>% group_by(Name, group) %>% slice(get_abs_max(tem)) %>% ungroup

#  Name    tem group var  
#  <chr> <dbl> <chr> <chr>
#1 Jon     -23 max   lan  
#2 Jon      63 min   sea  
#3 Maria    32 max   lan  
#4 Maria    58 min   lan  
#5 dan      55 max   sea  
#6 dan      66 min   sea

Or the same function in data.table -

library(data.table)

setDT(df)[, .SD[get_abs_max(tem)], .(Name, group)]

CodePudding user response：

df |> mutate(abs_tem = abs(tem)) |> 
      group_by(Name,group) |>  
      dplyr::filter(abs_tem == max(abs_tem)) |> 
      select(-abs_tem)

# A tibble: 6 × 4
# Groups:   Name, group [6]
  Name    tem group var  
  <chr> <dbl> <chr> <chr>
1 Jon     -23 max   lan  
2 Maria    32 max   lan  
3 Maria    58 min   lan  
4 Jon      63 min   sea  
5 dan      55 max   sea  
6 dan      66 min   sea