jiapingz/shapedtw

A shape-embedded dynamic time warping (DTW) algorithm. To the best of our knowledge, shapeDTW beats all other DTW variants on UCR time series datasets.

shapedtw's Issues

License?

What license is this code released under? I would like to create a version of it in Julia for a project.

Do you have python Code implementation of this ?

output type

When I implement this method on two-dimension data,why the calculated distance is two columns？
For example when I input
data1:118.9158 38.7927
118.9147 38.7928
data2:118.9158 38.7927
118.9147 38.7928
output: 0 0
why the result is two-dimension instead of 0?

Question towards Implementation

Hey,

i read your paper about shapeDTW and its implementation. You did really nice work there.
Can you explain if there is adifference between shapeDTW with Derivative shape descriptors with subsequences of lenght 3 and derivatie DTW? Because you also mentioned that you follow their forumla to compute numeric derivatives.
Furthermore, i would like to know if you choose a subsquence lenght of 10 for example, how do you comupte the distance between two subsequences (e.g. euclidean distance?...)? and you are storing the distance of the subsequences then back to gain a descriptor sequence to perform the DTW alignment right?

What module is "visualizer"?