The principal component analysis (PCA) technique, commonly used for complexity reduction in machine learning, may be inconsistent and unreliable, potentially contributing to the reproducibility crisis in science, according to a study. The misuse of machine learning and uncurated data were cited as reasons for the crisis, with genetics singled out as a field heavily reliant on PCA.

source update: Bioinformaticians Favorite Tool Can Be Misleading – Towards AI