R loop for univariate rolling window on multiple variables in DF - r

Question

I'm sure this is obvious but i'm a begginer in R and i spent a good part of the morning trying to solve this...
So, I have a DF of 24 time series of 3410 observations. I want to create a loop to procede a rolling window regression. I want to regress all the independant variable DF[,2:24] on the dependant variable DF[,1] in a rolling window and extract the evolution of the R2 in time.
I tried the following code but it isn't working:
resultsList <- list()
for (i in 1:ncol(var)) {
lmfit <- roll_regres(var[,1]~var[,i], var, 126, do_compute = c("sigmas", "r.squareds"))
resultsList[[i]] <- summary(lmfit)
}
I get the following error:
Error in roll_cpp(Y = y, X = x, window = width, do_compute_R_sqs = do_compute_R_sqs, :
'dchdd' failed with code 1
I also tried the following code using rollapply but it isn't working either:
resultsList <- list()
for (i in 1:ncol(var)) {
lmfit <- rollapply(var, width = 126, FUN = function(x) lm(var[, i] ~ var[,1]), by =1, align = "left")
resultsList[[i]] <- summary(lmfit)
}
I get the following error:
Error in zoo(rval, index(x)[i]) :
“x” : attempt to define invalid zoo object
If anyone could help me with this i would be VERY gratefull.
Thank you in advance for any help you can provide.

Answer 1

I solved the problem; here is the code:
#Loop RW 6 months
resultsList6M <- list()
for (i in 1:ncol(var)) {
lmfit <- roll_regres(var[,i]~cu, var, 126,do_compute = c("r.squareds"))
resultsList6M[[i]] <- lmfit["r.squareds"]
}
RS16m <- ts(as.numeric(as.character(unlist(resultsList6M[[1]]["r.squareds"]))),start = c(2007, 5), frequency = 252)
ts_plot(RS16m,
title = "126 Days RW, CADUSD ~ EURUSD ",
Xtitle = "Time",
Ytitle = "RSQUARED",
width = 1)
Now, I can do a graphical analysis of the R2 in time, i plan the repeat the rolling window on 6 months, 1 year and 3 years.