Practical Applications of Machine Learning for Econometricians: The Readme Manual

In Machine Learning for Econometricians: The Readme Manual, from the Summer 2022 issue of The Journal of Financial Data Science, Marcos López de Prado, of Cornell University and the Abu Dhabi Investment Authority, concisely covers many machine learning (ML) techniques and links them to analogous steps in the econometric research process: goal setting, outlier detection, visualization, feature extraction, regression, classification, feature importance, model selection, and validation. It is a must-read for econometricians who want to utilize increasingly available “big data” and the algorithms designed to process it. Econometricians should not be deterred by notions of ML as a black box. López de Prado explains that there are several methods for interpreting model results. ML consists of powerful techniques that can enrich our understanding of complex relationships that relatively simple traditional econometric methods cannot grasp. He does not imply that ML should replace econometrics but rather that it complements traditional methods of analysis. Furthermore, he notes that ignoring economic theory can result in poorly designed research studies and false discoveries. Importantly, he makes a strong case for econometricians to modernize their quantitative toolbox, and he offers a roadmap for updating their tools and improving their research skills.