fc_vine: Initialize a probabilistic power forecast for a specific time...

In kdayday/forecasting: Probabilistic Solar Forecsting

Description

Initialize a probabilistic power forecast for a specific time point, using an n-dimensional vine copula. Assumes training data already captures differences in magnitude (i.e., power rating) amongst sites. The "copula" field of the data.input list can be a matrix of training data [ntrain x nsites] OR a pre-trained vinecop model The "marginals" field of the data.input list can be a matrix of training data [ntrain x nsites] OR a list of single-site prob_forecast objects

Usage

fc_vine(data.input, location, time,
  training_transform_type = "empirical",
  results_transform_type = "empirical", n = 10000, samples.u = NA,
  ...)

Arguments

data.input	A list of "copula" and "marginals" inputs
location	A string
time	A lubridate time stamp
training_transform_type	Transform of training data into uniform domain (see marg_transform "cdf.method")
results_transform_type	Transform of copula results back into variable domain (see marg_transform "cdf.method")
n	An integer, number of copula samples to take
samples.u	(optional) A precalculated set of n-dimensional CDF samples from rvinecop
...	optional arguments to the marginal estimator

Value

An n-dimensional probabilistic forecast object from vine copulas

kdayday/forecasting documentation built on Oct. 7, 2020, 7:16 p.m.