What is Partial Least Squares (PLS)?

PLS is a supervised dimensionality reduction method maximizing covariance between features and target.

PLS finds latent components that explain both predictors and response variable, making it powerful for predictive modeling in high-dimensional settings.