| Title: | Nixtla's KAN Time Series Model In R |
|---|---|
| Description: | This package is a binding between Nixtla's neuralforecast Library, specifically KANs, and R's {modeltime} package. Nixtla's KAN is bound using {reticulate}, which is then ported into {parsnip} and bridged to {modeltime}. |
| Authors: | Frankie T. Hull [aut, cre] |
| Maintainer: | Frankie T. Hull <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 0.0.0.1 |
| Built: | 2025-02-21 06:18:19 UTC |
| Source: | https://github.com/frankiethull/kantime |
Create Virtual Environment Wrapper
create_neuralforecast_env(envname = "nixtla-dev", ...)create_neuralforecast_env(envname = "nixtla-dev", ...)
envname |
virtual environment to create |
... |
additional passes for |
creation of virtual environment
Install Development Version of neuralforecast From Github dev version contains conformal methods for predicting
install_neuralforecast(envname = "nixtla-dev", method = "auto", ...)install_neuralforecast(envname = "nixtla-dev", method = "auto", ...)
envname |
virtual environment name |
method |
method defaults to "auto" |
... |
additional passes for |
worth the wait! installs neuralforecast to "nixtla-dev" by default
Use Virtual Environment Wrapper
use_neuralforecast_env(envname = "nixtla-dev", ...)use_neuralforecast_env(envname = "nixtla-dev", ...)
envname |
virtual environment to use |
... |
additional passes for |
sets environment to virtualenv