wnw_pdf: Weighted Nadaraya Watson Estimator of Coditional PDF

In ygeunkim/ceshat: Nonparametric Estimation of Conditional Expected Shortfall

Description

This function estimates conditional pdf using WDKLL method.

Usage

wnw_pdf(
  formula,
  data,
  wt,
  nw_kernel = c("Gaussian", "Epanechinikov", "Tricube", "Boxcar"),
  nw_h,
  h0,
  init = 0,
  eps = 1e-05,
  iter = 1000
)

Arguments

formula	an object class formula.
data	an optional data to be used.
wt	weights for WNW. Computing in prediction step will help efficiency.
nw_kernel	Kernel for weighted nadaraya watson
nw_h	Bandwidth for WNW
h0	Binwidth
init	initial value for finding lambda
eps	small value
iter	maximum iteration when finding lambda

Details

Since standalone LL or WNW does not fully satisfy the conditions of cdf, Cai et al (2008) proposed to use WNW in LL scheme.

\hat{f}_c(y \mid x) = ∑_{t = 1}^n W_{c,t}(w, h) K_{h_0}(y - Y_t)

Value

Conditional pdf function of (y, x). y can be a numeric vector.

References

Cai, Z., & Wang, X. (2008). Nonparametric estimation of conditional VaR and expected shortfall. Journal of Econometrics, 147(1), 120-130.

ygeunkim/ceshat documentation built on Dec. 16, 2019, 12:39 p.m.