Learn R Programming

⚠️There's a newer version (2.1-10) of this package.Take me there.

stepR (version 2.1-4)

Multiscale Change-Point Inference

Description

Allows fitting of step-functions to univariate serial data where neither the number of jumps nor their positions is known by implementing the multiscale regression estimators SMUCE (K. Frick, A. Munk and H. Sieling, 2014) and HSMUCE (F. Pein, H. Sieling and A. Munk, 2017) . In addition, confidence intervals for the change-point locations and bands for the unknown signal can be obtained.

Copy Link

Version

Install

install.packages('stepR')

Monthly Downloads

233

Version

2.1-4

License

GPL-3

Maintainer

Pein Florian

Last Published

January 30th, 2023

Functions in stepR (2.1-4)

BesselPolynomial

Bessel Polynomials

Values of the MRC statistic for 1,000 observations (all intervals)

MRC.asymptotic.dyadic

"Asymptotic" values of the MRC statistic (dyadic intervals)

Bounds based on MRC

Computation of the multiscale statistic

Compute Multiresolution Criterion

Computation of the bounds

Reconstruct filtered piecewise constant functions with noise

Family of distributions

Critical values

monteCarloSimulation

Monte Carlo simulation

Confidence intervals for jumps and confidence bands for step functions

parametricFamily

Parametric families

Neighbouring integers

Digital filters

Interval systems

Multiscale Change-Point Inference

Piecewise constant regression with SMUCE

Fitted step function

Piecewise constant multiscale inference

Robust standard deviation estimate

Forward selection of candidate jumps

Automatic selection of number of jumps

Solution path of step-functions

Jump estimation under restrictions

testSmallScales

Test Small Scales

TRANSIT algorithm for detecting jumps

Compare fit blockwise with ground truth

"Asymptotic" values of the MRC statistic (all intervals)

Continuous time Markov chain