Dear R community,
I would like to introduce the package "dtw" for computing Dynamic Time
Warping (DTW) alignments and related algorithms on timeseries.
The DTW algorithm computes the time axis stretch which optimally maps one
timeseries (query) onto another (template); it outputs the resulting
cumulative distance between the two inputs, which is frequently used for
classification and clustering.
The R implementation in package "dtw" is fairly comprehensive,
allowing
choice of:
* windowing functions (global constraints), eg. the Sakoe-Chiba band
* 26 transition types (local constraints), incl.
- symmetric and asymmetric
- smoothed
- slope-constrained
- step patterns found in Sakoe-Chiba and Rabiner-Myers;
* any definition for a local distance can be used for multivariate
timeseries, thanks to the {proxy}distance function.
Besides alignment computations, the package also provides:
* plot methods for warping functions with several visualization styles;
* methods for applying a warping function (direct or inverse);
* native and interpreted implementations.
DTW itself registers as a distance function with {proxy}distance;
Package "dtw" is hosted on CRAN (stable) and R-forge (testing:
recommended).
Current package version is 1.4-3 (this is the first announcement on
r-packages).
Please find more information at the project home page:
http://dtw.r-forge.r-project.org , providing repository locations and a
gallery of sample plots. A companion paper is upcoming.
Best regards,
Toni Giorgino
--
Toni Giorgino, PhD toni.giorgino at unipv.it
Laboratorio di Informatica Biomedica tel +39 0382 985981
Dip. Informatica e Sistemistica fax +39 0382 526088
via Ferrata 1 I-27100 Pavia
[[alternative HTML version deleted]]
_______________________________________________
R-packages mailing list
R-packages at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-packages
