Trevor Hastie & Rob Tibshirani (along with Sourav Chatterjee) have just published a new paper “Univariate-guided sparse regression” where they present a new statistical tool called “uniLasso” that improves how we build predictive models from data. This method is especially useful when trying to understand which factors are truly important in complex data sets—a task that arises in fields ranging from health research to economics. https://hdsr.mitpress.mit.edu/pub/3i97j340/release/3
