In Dimensionality Reduction ML Techniques, what are the differences between the following 2 techniques—

a) PCA (Principal Component Analysis) &

b) Factor Analysis?

Good question! Both of them help in minimizing information loss, and have a few similar features. However, they are fundamentally different. PCA removes dependency/redundancy by removing those features which contain same information. In PCA, the derived components are independent of each other. The fundamental logic behind FA is to unearth the factors that store a variable’s spread. We also do FA to reduce the larger number of attributes. However, there are several factors that makes them different. For example, PCA is used to reduce the data into smaller number of components and so is a form of SVD. FA does a Common Factor Analysis to understand the underlying reason which these factors capture much of the information of a set of variables in the dataset data. PCA tries to identify dimensions that are composites of the observed predictors. FA explicitly presumes that the factors exist in the given data. You can find more such differences by searching on Google.

Rajtilak thanks for enlisting the differences between the two ML Reduction techniques i.e. PCA & Factor Analysis!!! It gives a better understanding regarding the two algorithms that are commonly used for Dimensionality Reduction. Your crisp explanation helps in understanding the differences between the two ML concepts… Indeed it does help!!!