itamarcaspi / psymonitor Goto Github PK

Hi!
I think there is a bug on line 77 in cvPSYwmboot
wn <- matrix(rnorm(1), nrow = Tb, ncol = nboot) repeats the same realization across the matrix, my interpretation of the paper is that something like:
wn <- matrix(rnorm(Tb*nboot), nrow = Tb, ncol = nboot)
would be more appropriate? Great package!

Hello Itamar,

I would like to raise three issues about the cvPSYmc.R function.

The first is that it seems to me that there is an error in the final calculation of the quantiles:

Q_PSY <- as.matrix(quantile(MPSY, qe), na.rm = TRUE)

The above calculation should not be done on the full MPSY data frame, but on the maximum of each line, like this:

aux <- apply(MPSY, MARGIN = 1, FUN = max, na.rm = T)
Q_PSY <- quantile(aux, qe)

Am I right?

The second issue is about the Tb variable that plays no role on the function.

And finally, this part of the code is not working properly:

if (useParallel == TRUE && missing(nCores)) { nCores <- detectCores() - 1 } else { nCores <- 1 }

I think that the problem is on the missing funtion.

Excuse me for any inaccuracy or error I eventually made. Thank you very much for the package!

Hi,

Thank you very much for sharing your code.

A very quick question. My input is discontinuous data that starts from 2017.12, but during 2018.01 - 2018.03, the price series is n/a due to some reason like no public trading.

The error message seems to be some issue relevant to the n/a in the data:

Error in betaM[[lag0]] :
attempt to select less than one element in get1index

The ideal output is something like this (done by GSADF test in Matlab):

Many thanks in advance!

Best wishes.

This code fails for count = 1, due to OT not being a matrix anymore.
Possible solution:
if (count == 1) {
OT <- matrix(EP[1:count, ], nrow=1, ncol=2)
} else {
OT <- EP[1:count, ]
}