Data scientists use dimensionality reduction in machine learning models to remove irrelevant features from busy datasets.