Abstract: To visualize or cluster high-dimensional data sets, it may be necessary to reduce the dimension. Principal component analysis (PCA) is one of the major linear methods to find a low-dimensional representation of a data set. In this lecture, we will introduce PCA from two different viewpoints: (1) by finding the best d-dimensional affine subspace to approximate the data, and (2) by finding the d-dimensional projection that preserves as much variance of the data as possible. We will conclude by showing that these interpretations are actually equivalent.
Principal component analysis for dimension reduction
15.03.2022 14:00 - 14:20
Organiser:
R. I. Boţ
Location:
HS 15, 2. OG, OMP1