Cluster time series

[jiedxu]

• about EM/GMM, CrossValidated

To Do

• [ ] read 3.3 of [liao2015clustering]
• [ ] why those for time series are different from ordinary methods.

[jiedxu]

This paper groups previously developed time series clustering methods into three major categories depending upon whether they work directly with raw data, indirectly with features extracted from the raw data, or indirectly with mod- els built from the raw data. [liao2015clustering]

[jiedxu]

Dynamic Time Warping (DTW)

DTW aligns the two series so that their difference is minimized. [liao2015clustering]

[jiedxu]

Model-based Approach

This class of approaches considers that each time series is generated by some kind of model or by a mixture of underlying probability distributions. Time series are considered similar when the models characterizing individual series or the remaining residuals after fitting the model are similar. [liao2015clustering]

• 81: Model-based clustering using Prophet

[jiedxu]

Proximity Measure

• [ ] Batista, G. E., Keogh, E. J., Tataw, O. M., & De Souza, V. M. (2014). CID: an efficient complexity-invariant distance for time series. Data Mining and Knowledge Discovery, 28(3), 634-669.
• [ ] 3.5 Proximity measures for structured data in [everitt2011cluster]