densratio {densratio} | R Documentation |
Estimate Density Ratio p(x)/q(x)
Description
Estimate Density Ratio p(x)/q(x)
Usage
densratio(x1, x2, method = c("uLSIF", "RuLSIF", "KLIEP"),
sigma = "auto", lambda = "auto", alpha = 0.1, kernel_num = 100,
fold = 5, verbose = TRUE)
Arguments
x1 |
numeric vector or matrix. Data from a numerator distribution p(x). |
x2 |
numeric vector or matrix. Data from a denominator distribution q(x). |
method |
"uLSIF" (default), "RuLSIF", or "KLIEP". |
sigma |
positive numeric vector. Search range of Gaussian kernel bandwidth. |
lambda |
positive numeric vector. Search range of regularization parameter for uLSIF and RuLSIF. |
alpha |
numeric in [0, 1]. Relative parameter for RuLSIF. Default 0.1. |
kernel_num |
positive integer. Number of kernels. |
fold |
positive integer. Numer of the folds of cross validation for KLIEP. |
verbose |
logical (default TRUE). |
Value
densratio object that contains a function to compute estimated density ratio.
Examples
x1 <- rnorm(200, mean = 1, sd = 1/8)
x2 <- rnorm(200, mean = 1, sd = 1/2)
densratio_obj <- densratio(x1, x2)
new_x <- seq(0, 2, by = 0.05)
estimated_density_ratio <- densratio_obj$compute_density_ratio(new_x)
plot(new_x, estimated_density_ratio, pch=19)
[Package densratio version 0.2.1 Index]