Dimensionality reduction is the process of discovering the explanatory variables that account for the greatest changes in the response variable.
The output is therefore similar to the input, but the number of features is reduced.
There are two main approaches.
We are looking for a projection that distorts the point cloud as little as possible. The dispersion of the data must be preserved as much as possible.
We model the variety on which the data are located (a d-dimensional variety is a part of a higher dimensional space).
Status:: #wiki/notes/mature
Plantations:: Data Science - 20230221095743
References::