Applying a function to a list and outputting results - r

Question

I've got a big database which I've split up by year and created at train and test for each respective year
#split the dataset into a list of datasets
Y <- split(dat_all, dat_all$year)
#create a train and test dataset for all years
#takes Y is inp
create_sets <- function(x){
train_set <- sample(2, nrow(x), replace = TRUE, prob = c(0.7, 0.3))
train <- x[train_set == 1, ]
test <- x[train_set == 2, ]
assign('x', list(train = train, test = test))
}
Ylist <- lapply(Y, create_sets)
To call each item out you use Ylist$'2016'$train
I've made an accuracy ratio function which I can run each list through individually but I am looking for a way to do it all in one to save massive amounts of code (theres 16 years of data)
Below is how I currently create an accuracy ratio for one year
val_train<-Ylist$'2016'$train
val_train$pred<-predict(modf,newdata=Ylist$'2016'$train)
val_train$probs<-exp(val_train$pred)/(1+exp(val_train$pred))
x<-data.frame(rcorr.cens(val_train$probs, val_train$default_flag))
train_AR<-x[2,1]
train_AR
modfull <-ModFit(test)
val_test<-test
val_test$pred<-predict(modf,newdata=test)
val_test$probs<-exp(val_test$pred)/(1+exp(val_test$pred))
x<-data.frame(rcorr.cens(val_test$probs, val_test$default_flag))
test_AR<-x[2,1]
test_AR
AR_Logistic1<-c(train_AR,test_AR,)
AR_Logistic2<-c(train_AR,test_AR) #just in to see if table works
AccuracyRatio<-rbind(AR_Logistic1,AR_Logistic2)
colnames(AccuracyRatio)<-c("Train","Test","All")
AccuracyRatio
Just to clarify I'm trying to run through my whole list through my accuracy ratio and then output the AR for each year for its train and test.
Any help is greatly appreciated

Answer 1

With lapply and wrapping the AR calculations in a function you can summarise the output as below.
Without sample data, I could not test it but let us know if you face any errors.
fn_Calc_AR <- function(yearDat = listInput) {
#yearDat <== Ylist$'2016'
trainDat <- yearDat$train
testDat <- yearDat$test
val_train<- trainDat
val_train$pred<-predict(modf,newdata= trainDat)
val_train$probs<-exp(val_train$pred)/(1+exp(val_train$pred))
x<-data.frame(rcorr.cens(val_train$probs, val_train$default_flag))
train_AR<-x[2,1]
#train_AR
modfull <-ModFit(testDat)
val_test<-testDat
val_test$pred<-predict(modf,newdata=testDat)
val_test$probs<-exp(val_test$pred)/(1+exp(val_test$pred))
x<-data.frame(rcorr.cens(val_test$probs, val_test$default_flag))
test_AR<-x[2,1]
#test_AR
AR_Logistic1<-c(train_AR,test_AR) # removed extraneous comma, previous input c(train_AR,test_AR,)
AR_Logistic2<-c(train_AR,test_AR) #just in to see if table works
AccuracyRatio<-rbind(AR_Logistic1,AR_Logistic2)
colnames(AccuracyRatio)<-c("Train","Test","All")
#confirm yearName is being created
try(yearName <- head(names(x),1)) #retain only year
if(length(yearName) > 0L) {
AR_DF <- data.frame(yearName = yearName , AccuracyRatio,stringsAsFactors=FALSE)
}else{
AR_DF <- AccuracyRatio
}
return(AR_DF)
}
Summarise Output:
AR_Summary = do.call(rbind,lapply(Ylist,fn_Calc_AR))
Aggregate Dataset:
aggregateTrain = do.call(rbind,lapply(Ylist,function(x) x$train))
aggregateTest = do.call(rbind,lapply(Ylist,function(x) x$test))
aggregateList = list(train = aggregateTrain,test = aggregateTest)
AR_AggregateSummary = do.call(rbind,lapply(aggregateList,function(x) fn_Calc_AR(x) ))