probtranskde: Get KDE using Geenens et. al 2014 and 2018's methods: 1)...

In kdayday/forecasting: Probabilistic Solar Forecsting

Description

Get KDE using Geenens et. al 2014 and 2018's methods: 1) transformation with probit or log function, 2) local likelihood estimation 3) nearest-neighbor bandwdith selection

Usage

probtranskde(x, xmax, scale = 0.9999, zero_offset = 1e-04,
  max_scaler = 2, weight = "WLSCV2", n.res = 500)

Arguments

x	A vector of samples
xmax	Maximum allowable x value for probit transformation, or NaN for log transformation (non-negative)
scale	Scaling factor (0,1] to move maximum values off boundary at 1
zero_offset	Amount to shift minimum values off the boundary at 0 in the (0,1) domain
max_scaler	For log transforms, estimation is made over the range max(x)*max_scaler
weight	One of 'LSCV', 'WLSCV1', 'WLSCV2'
n.res	length of resulting estimation vector

Value

A list of the estimate evaluation points, density, and cumulative distribution

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