Bold cells in kable (format = latex) based on rowwise largest value by group of variables

Working Example

set.seed(1234)
df <- data.frame(  x1 = rnorm(3), x2 = rnorm(3), z1 =rnorm(3), z2=rnorm(3))

which produces

 x1       x2       z1       z2
-------  -------  -------  -------
 -1.207   -2.346   -0.575   -0.890
  0.277    0.429   -0.547   -0.477
  1.084    0.506   -0.564   -0.998

I can then export to latex using

kable(df,format = "latex", digits = 3)

But I would like to have in bold the cells that are the largest across rows (x1 vs x2, and z1 vs z2). So in this example, x1=-1.207 and z1=-0.575 would be in bold in the first row, x2=0.429 and z2=-0.477 would be in bold in the second row, etc.

So that the final output should be

\begin{tabular}{r|r|r|r}
\hline
x1 & x2 & z1 & z2\\
\hline
\textbf{-1.207} & -2.346 & \textbf{-0.575} & -0.890\\
\hline
0.277 & \textbf{0.429} & -0.547 & \textbf{-0.477}\\
\hline
\textbf{1.084} & 0.506 & \textbf{-0.564} & -0.998\\
\hline
\end{tabular}

Any help would be great!

Solution

Using cell conditional formatting from kableExtra coupled with tidyverse packages this is one attempt, although I'd be the first to admit it seems alot of work.

I'm sure there is a better way.

```{r results='asis'}

library(dplyr)
library(tidyr)
library(tibble)
library(stringr)
library(kableExtra)

set.seed(1234)

df <-
  data.frame( x1 = rnorm(3), x2 = rnorm(3), z1 = rnorm(3), z2 = rnorm(3)) %>%
  rowid_to_column() %>%
  pivot_longer(-rowid)%>%
  mutate(group = str_sub(name, 1, 1),
  id = str_sub(name, -1, -1)) %>%
  group_by(group, rowid) %>%
  mutate(value = round(value, 3),
  value = ifelse(value == max(value), cell_spec(sprintf("%.3f", value), "latex", bold = TRUE),
                 sprintf("%.3f", value))) %>%
  ungroup() %>%
  select(-c(group, id)) %>%
  pivot_wider() %>% 
  select(-rowid)


kbl(df,
    booktabs = TRUE,
    escape = FALSE,
    align = "r")
```