R loop through multiple sub groups with using functions - r

Question

Hi I am trying to learn how to loop through multiple groups within a data frame and apply certain arithmetic operations. I do not have a programming background and am struggling to loop through the multiple conditions.
My data looks like the following:
Event = c(1,1,1,1,1,2,2,2,2,2)
Indiv1=c(4,5,6,11,45,66,8,9,32,45)
Indiv2=c(7,81,91,67,12,34,56,78,90,12)
Category=c(1,1,2,2,2,1,2,2,1,1)
Play_together=c(1,0,1,1,1,1,1,1,0,1)
Money=c(23,11,78,-9,-12,345,09,43,21,90)
z = data.frame(Event,Indiv1,Indiv2,Category,Play_together,Money)
What I would like to do is to look through each event and each category and take the average value of Money in cases where Play_together == 1. When Play_together==0, then I would like to apply Money/100.
I understand that the loop would look something like the following:
for i in 1:nrow(z){
#loop for event{
#loop for Category{
#Define avg or division function
}
}
}
However, I cannot seem to implement this using a nested loop. I saw another post (link: apply function for each subgroup) which uses dplyr package. I was wondering if someone could help me to implement this without using any packages (I know this might take longer as compared to using R packages). I am trying to learn R and this is the first time I am working with nested loops.
The final output will look like this:
where for event 1, the following holds:
a) For cateory 1:
Play_together ==1 in row 1; we take the avg of Money value and hence final output = 23/1= 23
Play_together==0 in row 2; we take Money/100= 0.11
b) For category 2:
Play_together == 1 for all observations. We take avg Money for all three observations.
This holds similarly for Event 2. In my actual dataset, I have event = 600 and number of category ranging from 1 - 10. Some events may have only 1 category and a maximum of 10 categories. So any function needs to be extremely flexible. The total number of observations in my dataset is around 1.5 million so any changes in the looping process to reduce the time taken to carry out the operation is going to be extremely helpful (Although at this stage my priority is the looping process itself).
Once again it would be a great help if you can show me how to use nested looping and explain the steps in brief. Much appreciated.

Answer 1

will something like this do?
I know it's using dplyr, but that package is made for this kind of jobs ;-)
Event = c(1,1,1,1,1,2,2,2,2,2)
Indiv1=c(4,5,6,11,45,66,8,9,32,45)
Indiv2=c(7,81,91,67,12,34,56,78,90,12)
Category=c(1,1,2,2,2,1,2,2,1,1)
Play_together=c(1,0,1,1,1,1,1,1,0,1)
Money=c(23,11,78,-9,-12,345,09,43,21,90)
z = data.frame(Event,Indiv1,Indiv2,Category,Play_together,Money)
library(dplyr)
df_temp <- z %>%
group_by( Event, Category, Play_together ) %>%
summarise( money_mean = mean( Money ) ) %>%
mutate( final_output = ifelse( Play_together == 0, money_mean / 100, money_mean )) %>%
select( -money_mean )
df <- z %>%
left_join(df_temp, by = c("Event", "Category", "Play_together" )) %>%
arrange(Event, Category)

Answer 2

Consider base R's by, the object-oriented wrapper to tapply designed to subset dataframes by factor(s) but unlike split can pass subsets into a defined function. Then, run conditional logic with ifelse for Final_Output field. Finally, stack all subsetted dataframes for final object.
# LIST OF DATAFRAMES
by_list <- by(z, z[c("Event", "Category")], function(sub) {
tmp <- subset(sub, Play_together==1)
sub$Final_Output <- ifelse(sub$Play_together == 1, mean(tmp$Money), sub$Money/100)
return(sub)
})
# APPEND ALL DATAFRAMES
final_df <- do.call(rbind, by_list)
row.names(final_df) <- NULL
final_df
# Event Indiv1 Indiv2 Category Play_together Money Final_Output
# 1 1 4 7 1 1 23 23.00
# 2 1 5 81 1 0 11 0.11
# 3 2 66 34 1 1 345 217.50
# 4 2 32 90 1 0 21 0.21
# 5 2 45 12 1 1 90 217.50
# 6 1 6 91 2 1 78 19.00
# 7 1 11 67 2 1 -9 19.00
# 8 1 45 12 2 1 -12 19.00
# 9 2 8 56 2 1 9 26.00
# 10 2 9 78 2 1 43 26.00