What is PCA?

Principal Components Analysis (PCA) is a simple algorithm often used for data analysis. Raw data collected from the real world can be very noisy or extraneous. PCA is a technique rooted in linear algebra that addresses this problem. At heart, it is a method of determining the directions of greatest importance in a dataset, using the data’s intrinsic statistics. PCA is most often used for dimensionality reduction, though it is commonly seen in data compression and visualization.

An illustration of PCA

This article introduces the concept of PCA, then moves into theory and knowledge, focusing on the exact steps of the algorithm. It then goes into the several common uses and applications of PCA, and finishes with a primer on limitations and extensions.