Academia's failure to retain data scientists
Francisco Rodríguez‐Sánchez, Ben Marwick, Edward D. Lazowska, Jake Vanderplas · Science · 2017
WE APPLAUD THE recommendations of V. Stodden et al. for “Enhancing reproducibility for computational methods” (Policy Forum, 9 December 2016, p. 1240). Many of their recommendations could be fulfilled if researchers embraced modern statistical and computational (“data science”) practices. However, academia has proven slow to adapt best practices and thus has failed to retain many brilliant quantitative scientists. This is bad for science, as prevalent data analyses are often suboptimal, data-driven discoveries remain underdeveloped, and highly skilled people, who could use their unique expertise to move forward the most pressing scientific questions, are lost to industry.