Loop Taking Too Long to Finish in R - r

Question

I have some code which includes a for loop, and nested if statements. The issue is that it is taking too long to run and I want to make it much faster.
I have data on cohorts in a data frame called f2_cebu_davao. There is also a column in this data frame called person_id. There are 3 categories of the cohorts: 'Baseline', 'Other Effects', 'Campaign'.
I want to loop through each person_id in the f2_cebu_davao data frame, and check to see which cohort it is in. If it is in the cohort 'Baseline' or 'Other Effects', then I will check the before_baseline_othereffects table to see if the ID can be found in that table. If it can, I make a new column in the f2_cebu_davao table and the value will be 'returning'. Otherwise, 'new'.
If the cohort name is 'campaign', I will check the before_campaign table and do the same procedure as above.
My data is quite big (all my objects are big) so this is taking a really long time to run (it's been running for more than 30 minutes and still not finished!).
How can I speed this up (possibly by using vectorization, or just by modifying the code a little)?
I tried do loop through but it's taking too long.
before_baseline_othereffects <- subset(loans_final_full, submitted_at_date < '2018-05-21')
before_campaign <- subset(loans_final_full, submitted_at_date < '2019-01-21')
for(i in 1:nrow(f2_cebu_davao)){
if(as.vector(f2_cebu_davao[, 'cohort'][i]) == 'Baseline') {
if(as.vector(f2_cebu_davao[,'person_id'][i]) %in% as.vector(unique(before_baseline_othereffects$person_id)) == TRUE) {
f2_cebu_davao$new_or_returning[i] <- 'Returning'
} else {
f2_cebu_davao$new_or_returning[i] <- 'New'
}
} else if (as.vector(f2_cebu_davao[, 'cohort'][i]) == 'Other Effects'){
if(as.vector(f2_cebu_davao[,'person_id'][i]) %in% as.vector(unique(before_baseline_othereffects$person_id)) == TRUE) {
f2_cebu_davao$new_or_returning[i] <- 'Returning'
} else {
f2_cebu_davao$new_or_returning[i] <- 'New'
}
} else {
if(as.vector(f2_cebu_davao[,'person_id'][i]) %in% as.vector(unique(before_campaign$person_id)) == TRUE) {
f2_cebu_davao$new_or_returning[i] <- 'Returning'
} else {
f2_cebu_davao$new_or_returning[i] <- 'New'
}
}
}

Answer 1

Happy to update and test this if you can provide some example data and desired output. I expect something like this should work.
Here I make up some fake data:
f2_cebu_davao <- data.frame(stringsAsFactors = F,
cohort = rep(c("Baseline", "Other Effects", "Something else",
"Another Something"), by = 3),
person_id = 1:12
)
before_baseline_othereffects <- c(1:4)
before_campaign <- c(5:8)
Here I apply it using dplyr's case_when, spelling out four cases. This code will be vectorized and I expect would run much faster than the current loop code.
The cohort is either "Baseline" or "Other Effects, and the person_id appears in before_baseline_othereffects. This creates "Returning" in rows 1 & 2.
Given the first condition wasn't met, but the cohort is still in either "Baseline" or "Other Effects," return "New," as is done in rows 5 & 6.
Given the first two conditions weren't met, but the person was in before_campaign, mark Returning, as in rows 7 & 8.
Otherwise, mark New, as in rows 3&4 and 9-12.
library(dplyr)
output <- f2_cebu_davao %>%
mutate(new_or_returning = case_when(
cohort %in% c("Baseline", "Other Effects") &
person_id %in% before_baseline_othereffects ~ "Returning",
cohort %in% c("Baseline", "Other Effects") ~ "New",
person_id %in% before_campaign ~ "Returning",
TRUE ~ "New"
))
Here's the output:
> output
cohort person_id new_or_returning
1 Baseline 1 Returning
2 Other Effects 2 Returning
3 Something else 3 New
4 Another Something 4 New
5 Baseline 5 New
6 Other Effects 6 New
7 Something else 7 Returning
8 Another Something 8 Returning
9 Baseline 9 New
10 Other Effects 10 New
11 Something else 11 New
12 Another Something 12 New