We are now done with the pre-processing of the data. It’s time to talk about dimension reduction.
We won’t go through the mathematical details, but instead aim for the intuition of how dimensional reduction methods (PCA, tSNE, UMAP) work. We want to learn how to reduce dimensions and visualise our data. We also learn how to select the principal components for the clustering step.
01:57 PCA
08:50 tSNE and UMAP for visualisation
10:05 tSNE
11:23 UMAP
Please note that the slide "tSNE simplified" is from StatQuest, which also provides an excellent video on how tSNE works ([ Ссылка ]).
We also recommend the excellent StatQuest videos explaining PCA, such as:
[ Ссылка ]
or:
[ Ссылка ]
Another comprehensive description of dimensionality reduction methods is a video by Paulo Czarnewski : [ Ссылка ].
Finally, the image on the slide "Other dimension reduction methods: used later for visualisation" is by Shigeo Takahashi, Issei Fujishiro, and Masato Okada, "Applying Manifold Learning to Plotting Approximate Contour Trees," IEEE Transactions on Visualization and Computer Graphics (Proceedings of IEEE Visualization / Information Visualization 2009), Vol. 15, No. 6, pp. 1185-1192, 2009.
