stepR (version 1.0-1)

stepR-package: Fit step-functions

Description

Allows to fit step-functions to univariate serial data where the number of jumps is used as a penalty.

encoding

utf8

Details

ll{ Package: stepR Version: 1.0 Date: 2013-09-17 Authors@R: c(person("Thomas", "Hotz", role = c("aut", "cre"), email = "[e-mail]"), person("Hannes", "Sieling", role = "aut", email = "[e-mail]")) Depends: R (>= 2.15.0), stats License: GPL-3 LazyData: yes Built: R 2.15.3; i686-pc-linux-gnu; 2013-09-17 07:05:08 UTC; unix }

Index: BesselPolynomial Bessel Polynomials MRC Compute Multiresolution Criterion MRC.1000 Values of the MRC statistic for 1,000 observations (all intervals) MRC.asymptotic "Asymptotic" values of the MRC statistic (all intervals) MRC.asymptotic.dyadic "Asymptotic" values of the MRC statistic (dyadic intervals) bounds Bounds based on MRC compareBlocks Compare fit blockwise with ground truth contMC Continuous time Markov chain dfilter Digital filters family Family of distributions jsmurf Reconstruct filtered piecewise constant functions with noise neighbours Neighbouring integers sdrobnorm Robust standard deviation estimate smuceR Piecewise constant regression with SMUCE stepR-package Fit step-functions stepblock Step function stepbound Jump estimation under restrictions stepcand Forward selection of candidate jumps stepfit Fitted step function steppath Solution path of step-functions stepsel Automatic selection of number of jumps transit TRANSIT algorithm for detecting jumps

References

Davies, P.L., Kovac, A. (2001). Local extremes, runs, strings and multiresolution. The Annals of Statistics 29, 1-65. Frick, K., Munk, A., and Sieling, H. (2014). Multiscale Change-Point Inference. With discussion and rejoinder by the authors. Journal of the Royal Statistical Society, Series B 76(3), 495-580. Friedrich, F., Kempe, A., Liebscher, V., Winkler, G. (2008). Complexity penalized M-estimation: fast computation. Journal of Computational and Graphical Statistics 17(1), 201-224. Futschik, A., Hotz, T., Munk, A. Sieling, H. (2013). Multiresolution DNA partitioning: statistical evidence for segments. Bioinformatics, online first: http://bioinformatics.oxfordjournals.org/content/early/2014/05/09/bioinformatics.btu180.full. Hotz, T., Schütte{Schuette}, O., Sieling, H., Polupanow, T., Diederichsen, U., Steinem, C., and Munk, A. (2013). Idealizing Ion Channel Recordings by a Jump Segmentation Multiresolution Filter. IEEE Transactions on NanoBioscience 12(4), 376-386. VanDongen, A.M.J. (1996). A New Algorithm for Idealizing Single Ion Channel Data Containing Multiple Unknown Conductance Levels. Biophysical Journal 70(3), 1303-1315.

See Also

smuceR, jsmurf, stepbound, steppath, stepsel, transit, compareBlocks, family

Examples

# estimating step-functions with Gaussian white noise added
# simulate a Gaussian hidden Markov model of length 1000 with 2 states
# with identical transition rates 0.01, and signal-to-noise ratio 2
sim <- contMC(1e3, 0:1, matrix(c(0, 0.01, 0.01, 0), 2), param=1/2)
plot(sim$data, cex = 0.1)
lines(sim$cont, col="red")
# maximum-likelihood estimation under multiresolution constraints
fit.MRC <- smuceR(sim$data$y, sim$data$x)
lines(fit.MRC, col="blue")
# choose number of jumps using BIC
path <- steppath(sim$data$y, sim$data$x, max.blocks=1e2)
fit.BIC <- path[[stepsel.BIC(path)]]
lines(fit.BIC, col="green3", lty = 2)

# estimate after filtering
# simulate filtered ion channel recording with two states
set.seed(9)
# sampling rate 10 kHz
sampling <- 1e4
# tenfold oversampling
over <- 10
# 1 kHz 4-pole Bessel-filter, adjusted for oversampling
cutoff <- 1e3
df.over <- dfilter("bessel", list(pole=4, cutoff=cutoff / sampling / over))
# two states, leaving state 1 at 10 Hz, state 2 at 20 Hz
rates <- rbind(c(0, 10), c(20, 0))
# simulate 0.5 s, level 0 corresponds to state 1, level 1 to state 2
# noise level is 0.3 after filtering
Sim <- contMC(0.5 * sampling, 0:1, rates, sampling=sampling, family="gaussKern",
  param = list(df=df.over, over=over, sd=0.3))
plot(Sim$data, pch = ".")
lines(Sim$discr, col = "red")
# fit under multiresolution constraints using filter corresponding to sample rate
df <- dfilter("bessel", list(pole=4, cutoff=cutoff / sampling))
Fit.MRC <- jsmurf(Sim$data$y, Sim$data$x, param=df, r=1e2)
lines(Fit.MRC, col = "blue")
# fit using TRANSIT
Fit.trans <- transit(Sim$data$y, Sim$data$x)
lines(Fit.trans, col = "green3", lty=2)