UMAP for time series decomposition?

[gabrielmpp]

Hey guys,

Are you aware of use cases of UMAP to decompose univariate time series? I was thinking if it would be possible to do something like the spectral decomposition analysis, that essentially applies PCA over a matrix where the columns are the same time series with different lags.

Thanks!

[EricKenjiLee]

There've been quite a few and I've listed two below. I've personally applied UMAP to a Toeplitz (lagged) matrix to form a sort of non-linear version of singular spectrum analysis (SSA) but I didn't pursue it very far. I'd be curious to know if you've found any new publications by now; I stopped looking a while back because I've been working on other things.

Ali, M., Jones, M. W., Xie, X. & Williams, M. TimeCluster: dimension reduction applied to temporal data for visual analytics. Vis Comput 35, 1013–1026 (2019).

DeAngelis, B. D., Zavatone-Veth, J. A. & Clark, D. A. The manifold structure of limb coordination in walking Drosophila. Elife 8, e46409 (2019).