In StevenGolovkine/denoisr: Nonparametric Kernel Smoothing Methods for Functional Data

knitr::opts_chunk$set(
  collapse=TRUE,
  comment = "#>",
  out.width = "100%")

denoisr

Overview

denoisr is a non-parametric smoother for noisy curve data, providing different functions to estimate various parameters:

• estimate_H0_list() and estimate_H0_deriv_list() estimate the smoothness of the curves.
• estimate_b_list() and estimate_bandwidth() estimate the bandwidth used in the Nadaraya-Watson estimator.
• estimate_curve() estimates one curve, given bandwidths.
• smooth_curves() and smooth_curves_regularity() estimate the curves.

You can learn more about them in vignette('denoisr').

Installation

To install the latest version directly from Github, please use

# install.packages("devtools")
devtools::install_github("StevenGolovkine/denoisr")

To build the vignette as well, please use

# install.packages("devtools")
devtools::install_github("StevenGolovkine/denoisr", build_vignettes = TRUE)

Dependencies

The denoisr package depends on the R-packages doParallel, dplyr, foreach, funData, iterators, KernSmooth, magrittr, np, parallel, purrr, Rcpp and RcppArmadillo.

References

The theoretical foundations of the estimation of regularity parameters and curves smoothing are described in:

Golovkine S., Klutchnikoff N., Patilea V. (2021) - Learning the smoothness of noisy curves with application to online curve reconstruction.

Bug reports

Please use GitHub issues for reporting bugs or issues.

StevenGolovkine/denoisr documentation built on Nov. 15, 2021, 8:44 a.m.