smooth
smooth copied to clipboard
config-i1
The set of functions used for time series analysis and in forecasting.
This should improve the functions in terms of speed, robustness and stability. 1. es() as a wrapper of adam() with distribution="dnorm"; 2. msarima() as a wrapper of adam() with model="NNN";...
There should be a mechanism of defining, which variables should be predicted and which should be used as is. A method preparing explanatory variables is needed. When the variable is...
This should call for adam() with distribution="plogis" and use a built-in mechanism of update of states. The main issue is with occurrence="general", which relies on a completely different stuff...
Develop a proper vcov for the three. Some parts of this do not work. An example: ```{r} require("smooth") set.seed(123) y
This is especially important in case of backcasting. Here is an example of very wrong initials: ```test
A few things to do: 1. Decide what this means for Etype=="M" and how it should be calculated; 2. Decide how to calculate them for the occurrence model; 3. Deal...
Explain, how the parameters are selected, when the default values are used (e.g. persistence, initial, sd etc). 1. sim.es 2. sim.ssarima 3. sim.ces 4. sim.sma 5. sim.gum 6. sim.ves
This can potentially done with something like vcov function. Currently we do not take the uncertainty of parameters into account.
Example. When bounds="a" in es() function, the optimal smoothing parameters differ from the case of bounds="u", and usually the value of CF in the former case is larger than in...