Looking at time series... if we non-dimensionalize and then try to compare them, looking for given patterns or for similarities that can be used to clasify them, how can we actually compare them? My thought was to discretize at different granularities (big slices, medium, and small slices) and then re-scan at each one, looking for patterns. Like taking music and trying to find a given three-note sequence, by first scanning it as a series of whole notes, then redoing it as all quarter notes, then sixteenths, etc., looking for the same three-note pattern at each level of detail. This would be used in classifying time-series Doppler-shifted astronomy data... seems reasonably simple to me.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
no subject
Date: 2002-09-27 03:15 am (UTC)