ftnonpar (version 0.1-88)

smqreg: Smooth piecewise monotone regression with taut strings

Description

Applies the smooth taut string method to one-dimensional data.

Usage

smqreg(y, thr.const=2.5, verbose=FALSE, bandwidth=-1, 
sigma=-1, localsqueezing=TRUE, squeezing.factor=0.5, DYADIC=TRUE,
firstlambda=100,smqeps=1/length(y),fsign=double(0),gensign=TRUE,
tolerance = 1e-12,...)

Arguments

y

observed values (ordered by value of independent variable)

thr.const

smoothing parameter for the multiresolution criterion (should be approximately 2.5)

verbose

logical, if T progress (for each iteration) is illustrated grahically

bandwidth

if set to a positive value the specified bandwidth is used instead of the multiresolution criterion.

sigma

if set to a positive value sigma the standard deviation is set to sigma and not estimated from the data

localsqueezing

logical, if T (default) the bandwidth is changed locally.

squeezing.factor

The amount of decrement applied to the bandwidthes

DYADIC

If TRUE the multiresolution constraints are only checked on dyadic intervals.

firstlambda

Initial value of lambda's for local or global squeezing.

smqeps

Distance between the (equally-spaced) time points.

fsign

Monotonicity constraints, vector of size n-1 of -1,0 and 1's. If fsign[i]==1, then fhat[i+1]>= fhat[i]. If fsign[i]==-1, then fhat[i+1]

gensign

If TRUE the taut string method is used to automatically produce suitable monotonicity constraints.

tolerance

Precision for the nested intervals for solving the minimisation problem.

...

Passed to the plot command if verbose=T.

Value

A list with components

y

The approximation of the given data

nmax

Number of local extreme values

sigma

Standard deviation used

References

Kovac, A. (2006) Smooth functions and local extreme values. Technical Report

See Also

pmreg,mintvmon,l1pmreg,pmden,pmspec

Examples

data(djdata)
par(mfrow=c(2,2))
plot(djblocks,col="grey")
lines(smqreg(djblocks)$y,col="red")
plot(djbumps,col="grey")
lines(smqreg(djbumps)$y,col="red")
plot(djheavisine,col="grey")
lines(smqreg(djheavisine)$y,col="red")
plot(djdoppler,col="grey")
lines(smqreg(djdoppler)$y,col="red")