I guess I could use Wilcoxon to test my time series against the climatic time series, with values in terms of mean+n*std or mean-n*std, where n could be 1, 2 o 3 for roughly 66, 95 or 99% confidence intervals for the climatic mean... But there may be a more elegant way to do this...

Thanks so much!

Hello,

I think you can use "Dynamic Time Warping" technique ,

See this reference:

https://www.aaai.org/Papers/Workshops/1994/WS-94-03/WS94-03-031.pdf

And there is the "dtw" R package: see the tutorial :

https://cran.r-project.org/web/packages/dtw/dtw.pdf

CV%(variability)= (St. dev / Mean) x 100

Regards

I agree with Bayan Hussien

Dear Francesc Peters,

The null hypothesis states that, Signed Wilcoxons test is a better choice, when the forecasting methods are resulting different forecasting accuracy.

But when the forecasting methods results same forecasting accuracy Diebold-Mariano (DM) statistical test is a better choice for it.

Thanks