How to perfom different operations for every row in one same column in a data frame

I would like to create an R function in which, the input data will be a data frame with the following structure:

And, for each different shape (i.e, 'Ellipsoid', 'Rectangular_box'), I would like to use a differente equation to calculate the volume using the respective value of the dimenstions (i.e, 'Dim_a', 'Dim_b', 'Dim_c').

For example, for 'Ellipsoid' shape, the equation to calculate the volume is:

vol = (pi/6) * Dim_a * Dim_b * Dim_c (eq. 1)

And for 'Rectangular_box', the equation is:

vol = Dim_a * Dim_b * Dim_c (eq. 2)

So, in my R function I would like to do "if Ellipsoid shape, then use eq. 1, but if Rectangular shape, then use eq. 2".

And the output is a new column with the results of calculate the volume for each different shape.

I was trying to do this:

Biovol3 <- function(data_frame){ #The input is a data frame
  
# The variables are: 'Shape' and the different dimentions 'Dim_a', 'Dim_b', 'Dim_c' that must be included in the data frame

  Shape <- data_frame$Shape
  Dim_a <- data_frame$Dim_a
  Dim_b <- data_frame$Dim_b
  Dim_c <- data_frame$Dim_c

# Then I tried to use 'which' function to select the shape
 
common_sp <- c("Ellipsoid", "Rectangular_box") # common shapes that must be included in the 'shape' column                 in the data frame

  sel_sp <- which(common_sp == Shape)

# Using 'if' statement to calculate the volume for each different shape
  
  if(any(sel_sp == 1)){
      
      vol = (pi/6) * Dim_a * Dim_b * Dim_c
    }
    
  if(any(sel_sp == 2)){
    
    vol = Dim_a * Dim_b * Dim_c
    }
  
# The output must be a data frame with a new column 'volume'
 
 result_data_frame <- data.frame(data_frame,
                                  vol = unname(vol),
                                  Area = unname(Area)) 
  
  return(result_data_frame)
} 

This returns me the following data frame as a result:

But the result of volume of Ellipsoid is incorrect. I notice that this is because the function only use one of the equations (in this case, the eq. 2), in both shapes, I don't know how to use the different equations corresponding to the different shapes.

in Base R you could do:

frac <- c(Ellipsoid = pi/6, Rectangular_box = 1)
df$vol <- frac[df$Shape] * df$Dim_a * df$Dim_b * df$Dim_c
df
            Shape Dim_a Dim_b Dim_c      vol
1       Ellipsoid    23    10    23 2769.838
2 Rectangular_box     4    65    18 4680.000

You could use mutate and ifelse

df %>% mutate (
        vol = ifelse(df$Shape %in% 'Ellipsoid',
            (pi/6) * Dim_a * Dim_b * Dim_c),
            Dim_a * Dim_b * Dim_c
    )

Dummy data:

library(tidyverse)
df1 <- tribble(~Shape,	~Dim_a,	~Dim_b,	~Dim_c,
               "Ellipsoid",	23,	10,	23,
               "Rectangular_box",	4,	65,	18)

You can create a list with your functions (with the same name as "shape"):

fun_list <- list(Ellipsoid = function(Dim_a, Dim_b, Dim_c) {(pi/6) * Dim_a * Dim_b * Dim_c},
                 Rectangular_box = function(Dim_a, Dim_b, Dim_c) {Dim_a * Dim_b * Dim_c})

And a list with every row of your dataframe as rows using purrr::transpose:

transpose(select(df1, c(Dim_a,	Dim_b,	Dim_c))

Now you can apply do.call to every combination of fun_list[df1$Shape], and the row list above, using purrr::map2:

map2_dbl(fun_list[df1$Shape], 
         transpose(select(df1, c(Dim_a,	Dim_b,	Dim_c))),
         do.call)

   Ellipsoid       Rectangular_box 
   2769.838        4680.000

For 2 options, ifelse is a great option. If you want something more extendable, case_when or case_match are better.

mutate, case_when, and case_match are all from the dplyr package, with case_match requiring a pretty recent version.

case_match is probably a bit cleaner if you're decision is based only on Shape, and you have a recent version of dplyr

df %>%
  mutate(
    vol = case_match(Shape,
       "Ellipsoid"       ~ (pi/6) * Dim_a * Dim_b * Dim_c,
       "Rectangular_box" ~ Dim_a * Dim_b * Dim_c,
       "Pyramid" ~ Dim_a * Dim_b * Dim_c / 3,
       .default ~ NA # default behavior, but modifiable here
    )
  )

case_when is more flexible if your conditions are based on more than one variable

df %>%
  mutate(
    vol = case_when(
       Shape == "Ellipsoid"       ~ (pi/6) * Dim_a * Dim_b * Dim_c,
       Shape == "Rectangular_box" ~ Dim_a * Dim_b * Dim_c,
       Shape == "Pyramid"         ~ Dim_a * Dim_b * Dim_c / 3,
       TRUE                       ~ NA  # default behavior, but modifiable here
    )
  )

Using either function, you can as many shapes as you need (e.g. the pyramid example I added)