Nested for-loops into apply? - r

Question

I need help figuring out how to improve a for loop. It doesn't necessarily needs to be apply, I just thought it was the best way after researching it on StackOverflow. I tried following the guides I found on StackOverflow, but I'm a newbie and believe I'm not getting a good grasp on the apply function.
This is a reconstruction of the snippet I'm trying to change:
for (j in 1:NROW(teste2) {
for (z in 1:NROW(teste2)) {
if(is.na(teste2$DATE[z]==Sys.Date()+j-1) | z==NROW(teste2)){
teste2$SALDO[z] <- 0
}else{
if(teste2$SALDO[z]!=0 & teste2$DATE[z]==Sys.Date()+j-1){
teste2$SALDO[z+1] <- teste2$PREVISAO_FINAL2[z] + teste2$SALDO[z+1]
}else{
teste2$SALDO[z] <- teste2$SALDO[z]
}
}
}
I tried doing the following:
for (j in 1:NROW(teste2) {
rows = 1:NROW(teste2)
saldo_fn <- function(z){
return(if(is.na(teste2$DATE[z]==Sys.Date()+j-1) | z==NROW(teste2)){
teste2$SALDO[z] <- 0
}else{
if(teste2$SALDO[z]!=0 & teste2$DATE[z]==Sys.Date()+j-1){
teste2$SALDO[z+1] <- teste2$PREVISAO_FINAL2[z] + teste2$SALDO[z+1]
}else{
teste2$SALDO[z] <- teste2$SALDO[z]
}
})
}
teste2$SALDO <- sapply(rows, saldo_fn)
}
But when I run sum(teste2$SALDO) it gives a different value.
What am I doing wrong? how do I fix it?

Answer 1

You cannot use apply-family function to optimize the algorithm. The reason is the line:
teste2$SALDO[z+1] <- teste2$PREVISAO_FINAL2[z] + teste2$SALDO[z+1]
You are recursively changing the value of next element based on the value of current one.
It is possible avoid for-loop in by using recursion, i.e. if you see something like x[i+1] = x[i+1] + x[i] you should use either for-loops or recursive functions (I prefer for-loops they are much easier and there is no problem with call stack overflow), if you see something like z[i] = F(x[i], y[i]), where F is some function, you can use apply-family functions.