The work of Dr. Eamonn Keogh at University of California Riverside has shown that a good way to classify time series is with a k-NN algorithm using a dynamic time warping similarity measure.
This repo is meant to implement this time series classification method in Python. The same techniques are also extended to clustering time series.
I have also written a tutorial on this subject.
