I have 20 time series each with different length and different time stamps (for example TS1 start 00:15 and ends 23:50 and TS2 start 00:30 and ends 23:00). I want to use dtw but i dont know how to motivate it.
I am thinking about using two time series and get ccf for each lag and plot the heatmap. But does ccf even interpretable when i have time series with different length? The hypothesis is that time series with similar surface structure as such as a high proportion of roads, are similar to each other.
My data looks something like this:
Area Time Day Value surface
1 00:30 1 0.2 50% roads
...
1 23:30 1 2.6 50% roads
......
2 00:15 1 2.7 48% roads
...
i did not find anything about the motivation of dtw using ccf of time series with different length