calculate difference with the previous element and having a defined starting point-CodePudding

Suppose we have the following df:

  df =  structure(list(fruit = c("melon", "mango", "orange", "blueberry"
    ), pct = c(5, 4, 3, 2)), class = "data.frame", row.names = c(NA, 
    -4L))

That looks like this:

      fruit pct
1     melon   5
2     mango   4
3    orange   3
4 blueberry   2

I want to start with a given number, say 30 and I would like to go across the column pct computing differences in this way:

     fruit pct desired_output
1     melon   5  30
2     mango   4  30-5=25
3    orange   3  25-4=21
4 blueberry   2  21-3=19

Note that the differences are being done with the previous element in the vector, and there is a starting point which in this case I defined as 30.

I have tried functions like diff and cumsum but I'm not getting the desired output.

CodePudding user response：

Happy new year! here is a slight variation

starting_point=30

df_new<-df %>%
  mutate(interim=lag(cumsum(pct))) %>%
  mutate(desired_output= starting_point-interim)

df_new$desired_output[1]=starting_point

CodePudding user response：

Please find below a slightly simpler solution with only base R

Code

x <- 30
df$desired_output <- replace(x - lag(cumsum(df$pct)), 1, x)

Output

df
#>      fruit pct desired_output
#>1     melon   5             30
#>2     mango   4             25
#>3    orange   3             21
#>4 blueberry   2             18

CodePudding user response：

Using data table:

Data:

df = structure(list(fruit = c("melon", "mango", "orange", "blueberry"
), pct = c(5, 4, 3, 2)), row.names = c(NA, -4L), class = c("data.table", 
"data.frame"), .internal.selfref = <pointer: 0x56485756c100>)

Code:

starting_point =  30
df[,cmsum:=starting_point-cumsum(c(0,pct[-length(pct)]))]