frimane/SolarClusGnr: Solar irradiance time-series clustering and downscaling

An unsupervised classification method using Dirichlet process Gaussian mixture model (DPGMM) is proposed in this package. The key benefit of the DPGMM paradigm is that it allows the ability to automatically adapt to the correct complexity level and size of the model, and can, therefore, fit complex probability functions. We apply a Markov chain Monte Carlo algorithm, namely Gibbs sampling for posterior inference. The package also deals with solar irradiance time-series downscaling. It can generate 1-min data worldwide.

install.packages("remotes")
remotes::install_github("frimane/SolarClusGnr")