Closed nipunbatra closed 11 years ago
is there an implementation out there? compatible license?
From dtw packages's page "The R Package dtw provides the most complete, freely-available (GPL) implementation of Dynamic Time Warping-type (DTW) algorithms up to date."
well - it could make sense to include but would have to have someone write a python impl (just to validate and get it right), then could cythonize to make it fast
@nipunreddevil up for it?
@jseabold is this more pandas or statsmodels?
@jreback How about first using Pandas-rpy2-R bridge and seeing how it works? What is the general policy regarding such things which exist open source in R?
Btw some more open source implementations here
http://mlpy.sourceforge.net/ Already implemented here - is there reason not to use this? (as opposed to writing another impl in pandas)
I posted some notes regarding the same here Even i am not in favor of reinventing the wheel!
scikit image has one of these iirc ;)
@sg: Thanks for pointing. scikit-image seems to have some stuff pertinent to time series
Just saw https://github.com/scikit-image/scikit-image/pull/518
But didn't find anything yet in documentation.
On Sun, Jun 9, 2013 at 12:51 PM, sg notifications@github.com wrote:
scikit image has one of these iirc ;)
— Reply to this email directly or view it on GitHubhttps://github.com/pydata/pandas/issues/3810#issuecomment-19162036 .
it's still pretty recent + is in review so it's not been merged - therefore nothing in the official docs
@mdeklerk @ahojnnes @stefanv
closing as not really pandas domain
R has a toolbox for it http://dtw.r-forge.r-project.org/
mlpy has something on these lines as well, though not as complete.
Since Pandas has a lot of timeseries stuff, adding DTW might be good. Or if not adding afresh, integrating with R or mlpy package maybe, or simply using rpy2 interface within Pandas and expand the sections here http://pandas.pydata.org/pandas-docs/dev/r_interface.html
References: