fittestArimaKF(CATS[,2],CATS.cont[,2])

[Lesaffrea]

HI,

using your example for fittestArimaKF as follow (from vignette)

data(CATS,CATS.cont)
fArimaKF <- fittestArimaKF(CATS[,2],CATS.cont[,2])
#predicted values
pred <- fArimaKF$pred

This function generate a warning NaNs produced and when you check the lower and upper of the return perditions you got NaN and the predict$Mean is totally wrong -45990427 -46084237 -46178173 -46272295 .

If use auto.arima one same data set I have got the issue, if I set stepwise=FALSE, approx=FALSE to avoid the short cut the upper and lower are calculated and no issue.

If I add those switch of fittestArimaKF I have still the issue mentioned above but the pred$mean is of -20223302 -20264685 -20306060 -20347432 -20388810 ...

I use forecast 8.12 and TSPred 4.00

Best regards

Alain