Stéphane Guerrier
Stéphane Guerrier
Robust WV CI
The CI for the WV are off. If the efficiency is 60% the robust CI at the 95% confidence level should be about 30% (100_( diff(c(-2,2)_sqrt(1/0.6))/diff(c(-2,2)) - 1)) wider then...
We need to be able to select scales to run the RGMWM. This should be done case by case. I think the default should be: (1) If model is stationary:...
Check with @robertomolinari for details
To estimate the parameters of stationary latent time series models with relatively small sample size (e.g. T < 1000) we need to implement the MLE of such models. This can...
Create a function to produce a graph similar to Fig 3 (right panel) of http://www.alexchinco.com/fast-trading-priced-noise.pdf
Create a new function for the WV to produce graphs similar to the one presented in p. 60 of Wavelet Applications in Economics and Finance edited by Mauro Gallegati, Willi...
Simulate ARMA(2,2) and compare: 1) MLE 2) GMWM exact 3) GMWM guess 4) Robust GMWM guess 5) Robust GMWM exact Produce boxplots. Replicate results for: 1) ARMA(2,2) with 1% contaminated...
It would be interesting to implement the Slepian Wavelet Variance which can be applied to irregular and gappy time series. See http://arxiv.org/pdf/1401.1785v1.pdf
- [ ] Add spatial functions (see folder `spatial-to-fix`) - [ ] Put uml somewhere else? - [ ] Function `robust_eda` there is a problem with the help (see examples)...
Hi guys, Once Issue #4 has been addressed the next part is to compute various statistics from the wavelet decomposition, here are the main ones: - [x] Classical wavelet variance:...