CODEWITHSUNDEEP
r
tadon11Aaa
tadon11Aaa

Reputation: 420

How can I renew my sample() in an ifelse statement?

I have data that looks like this.

   investor_name funding_round_type count
   <chr>         <chr>              <int>
 1 .406 Ventures angel                  1
 2 .406 Ventures other                  2
 3 .406 Ventures private-equity         1
 4 .406 Ventures series-a               5
 5 .406 Ventures series-b               2
 6 .406 Ventures series-c+              7
 7 .406 Ventures venture                1
 8 500 Startups  angel                 40

I would like to replace all the instances where funding_round_type is equal to venture and replace it with either series-a, series-b or series-c+. I'd like to randomly select one of those with a 40% chance for either of the first two and a 20% for the last one.

my_df %>% 
   mutate(funding_round_type = ifelse(funding_round_type == "venture", 
                                      sample(c("series-a", "series-b", "series-c"), 1, replace = TRUE, prob = c(.4, .4, .2)),
                                      funding_round_type))

Weirdly, the sample() seems to choose once and then revert to the chosen value for every row. I've run this a few times and it replaces venture with only one of the values from my list of options and doesn't include any instances of any of the other values.

How can I get sample() to run fresh on every row?

Upvotes: 2

Views: 48

Answers (3)

akrun
akrun

Reputation: 887311

We can use data.table methods

library(data.table)
setDT(df)[funding_round_type == "venture", funding_round_type := 
    sample(c("series-a", "series-b", "series-c+"), 1, prob = c(.4, .4, .2))][]
#    investor_name funding_round_type count
#1: .406 Ventures              angel     1
#2: .406 Ventures              other     2
#3: .406 Ventures     private-equity     1
#4: .406 Ventures           series-a     5
#5: .406 Ventures           series-b     2
#6: .406 Ventures          series-c+     7
#7: .406 Ventures           series-b     1
#8:  500 Startups              angel    40

Or using case_when from tidyverse

library(tidyerse)
df %>%
     mutate(funding_round_type = case_when(funding_round_type == "venture" ~ 
        sample(c("series-a", "series-b", "series-c+"), 1, prob = c(.4, .4, .2)),
       TRUE ~ funding_round_type))
#  investor_name funding_round_type count
#1 .406 Ventures              angel     1
#2 .406 Ventures              other     2
#3 .406 Ventures     private-equity     1
#4 .406 Ventures           series-a     5
#5 .406 Ventures           series-b     2
#6 .406 Ventures          series-c+     7
#7 .406 Ventures           series-a     1
#8  500 Startups              angel    40

data

df <- structure(list(investor_name = c(".406 Ventures", ".406 Ventures", 
".406 Ventures", ".406 Ventures", ".406 Ventures", ".406 Ventures", 
".406 Ventures", "500 Startups"), funding_round_type = c("angel", 
"other", "private-equity", "series-a", "series-b", "series-c+", 
"venture", "angel"), count = c(1L, 2L, 1L, 5L, 2L, 7L, 1L, 40L
)), class = "data.frame", row.names = c("1", "2", "3", "4", "5", 
"6", "7", "8"))

Upvotes: 0

Ronak Shah
Ronak Shah

Reputation: 389055

It is because ifelse runs the sample function only once and you are selecting one value from it which is recycled for every other value. Try doing

library(dplyr)

my_df %>% 
  mutate(funding_round_type = ifelse(funding_round_type == "venture", 
        sample(c("series-a", "series-b", "series-c"), 
        sum(funding_round_type == "venture"),replace = TRUE, prob = c(.4, .4, .2)), 
        funding_round_type))

Or with replace

my_df %>% 
   mutate(funding_round_type = replace(funding_round_type, 
   funding_round_type == "venture", sample(c("series-a", "series-b", "series-c"), 
   sum(funding_round_type == "venture"), replace = TRUE, prob = c(.4, .4, .2))))

Also you can replace this directly, without any ifelse or any packages.

my_df$funding_round_type[my_df$funding_round_type == "venture"] <-  
    with(my_df, sample(c("series-a", "series-b", "series-c"), 
    sum(funding_round_type == "venture"), replace = TRUE, prob = c(.4, .4, .2)))

Upvotes: 2

andrew_reece
andrew_reece

Reputation: 21274

Using rowwise() will resample for each row:

df %>%
  rowwise %>%
  mutate(funding_round_type = if_else(
    funding_round_type == "venture",
    sample(c("series-a", "series-b", "series-c+"), 1, prob = c(.4, .4, .2)),
    funding_round_type))

Also - minor, but you don't need replace=TRUE since you're only pulling one sample per call to sample().

Upvotes: 0

Related Questions