问题
This is a continuation of this question here: For-Loop By Columns with existing For-loop by Rows
I have a dataset in which I am using 3 variables:
adstock_rate
, diminishing_rate
, and lag_number
.
These are currently set to only 1 number each.
Currently I am using the following numbers:
adstock_rate<-0.5
lag_number<-1
diminishing_rate<-0.6
The final output is a dataset with new columns attached to the existing dataset using the function below.
foo1 <- function(dot, lag_val = 1) {
tmp <- dot
for(i in (1 + lag_val): length(tmp)) {
tmp[i] <- tmp[i] + adstock_rate * diminishing_rate * tmp[i - lag_val]
}
return(tmp)
}
advertising_dataset %>%
group_by(Region) %>%
mutate_all(funs(adstocked = foo1(., lag_val = lag_number)))
Here is what I am trying to do:
I want to apply this function to different values to these variables. Below are the combinations of these variables:
adstock_rate = c(0.50, 0.60, 0.70)
lag_number = c(0,1)
diminishing_rate = c(0.50, 0.60)
combos<-expand.grid(adstock_rate,lag_number,diminishing_rate)
colnames(combos)[1]<-"AdStock_Rate"
colnames(combos)[2]<-"Lag_Number"
colnames(combos)[3]<-"Diminish_Rate"
head(combos)
AdStock_Rate Lag_Number Diminish_Rate
1 0.5 0 0.5
2 0.6 0 0.5
3 0.7 0 0.5
4 0.5 1 0.5
5 0.6 1 0.5
6 0.7 1 0.5
7 0.5 0 0.6
8 0.6 0 0.6
9 0.7 0 0.6
10 0.5 1 0.6
I think you would have to make a for-loop or use the apply function to go down the list of rows in the combos dataset.
Here is my attempt:
for(j in combos){
foo1 <- function(dot, lag_val = 1) {
tmp <- dot
for(i in (1 + lag_val): length(tmp)) {
tmp[i] <- tmp[i] + combos[j,1] * combos[j,3] * tmp[i - lag_val]
}
return(tmp)
}
advertising_dataset %>%
group_by(Region) %>%
mutate_all(funs(adstocked = foo1(., lag_val = combos[j,2])))
##cbind to previous output
}
I also need the column names to have the number values such as adstock_0.5_1_0.6
where 0.5 = adstock rate, 1 = lag number, and diminishing = 0.6.
Hope this makes sense.
Please let me know if you need me to provide any more info.
Thanks!
回答1:
As we are looping through the rows of 'combos', create a list
that have the same length
as the number of rows of the 'combos' for storing the output from the for
loop
lst <- vector("list", nrow(combos)) # initialize a list to store output
Add some more parameters in the 'foo1' for more flexibility
foo1 <- function(dot, lag_val = 1, combos, ind) {
tmp <- dot
for(i in (1 + lag_val): length(tmp)) {
tmp[i] <- tmp[i] + combos[ind,1] * combos[ind,3] * tmp[i - lag_val]
}
return(tmp)
}
and then loop through the rows of 'combos'
for(j in seq_len(nrow(combos))){
# assign the group by recursive output to each `list` element
lst[[j]] <- advertising_dataset %>%
group_by(Region) %>%
mutate_all(funs(adstocked =
foo1(., lag_val = combos[j,2], combos, ind = j)))
}
lst
It is not clear whether we need the list
names to be 'adstock_Rate_Number_Drate' or not. If that is that case,
names(lst) <- paste0("adstock_", do.call(paste, c(combos, sep="_")))
Convert the list
of data.frame
to a single data.frame having an 'id' column to indicate the combination
out <- bind_rows(lst, .id = 'id')
head(out, 3)
# A tibble: 3 x 6
# Groups: Region [1]
# id Region advertising advertising2 advertising_adst… advertising2_ads…
# <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
#1 adstock_… 500 118. 43.9 147. 54.9
#2 adstock_… 500 120. 231. 150. 289.
#3 adstock_… 500 126. 76.8 157. 96.0
回答2:
To add on to @akrun's answer. If we wanted to have it in column form this is how you would do it. @akrun if you think there is a better way, let me know:
test<-out %>%
gather(var, value, -(id:Region)) %>%
unite(var, var, id, sep="_") %>%
spread(var, value)
colnames(test)
colnames(test) = gsub("_adstock_", "+", colnames(test))
colnames(test) = gsub("^(?!.*adstocked)([^+]*)\\+.*","\\1", colnames(test), perl=TRUE)
non_dupe<-test[!duplicated(as.list(test))]
来源:https://stackoverflow.com/questions/50256439/multiple-variable-values-to-function-and-cbind-results