\pretitle\posttitle\preauthor\postauthor\predate\postdate
end rewrite
The capacity of machine learning algorithms to predict certain outcomes with a high level of accuracy, and reveal non-linear relationships among variables are considered the main advantages of these methods over conventional assessments made by experts and inferential studies based on linear regression \cite{berk2014forecasts, athey2015machine, varian2014big}.
Furthermore, some researchers argue that these predictive methods provide a solid and superior alternative to both clinical or expert assessments in the case of practitioners and linear regression models in criminology research, where data available typically compromise the basic assumptions that produce desirable statistical properties for causal inference.
rewrite
No matter the predictive merits of statistical learning argued by researchers and practitioners, there are as many proponents as detractors in both the academic and the practitioner side of their use in policy. They are a pink elephant in the room of the many criminal justice institutions at the state and federal levels currently using predictive tools.