Fundamentals folks. A great example is the paper on police misconduct. It highlights a lot of great data science practices (more than I could squeeze into the video). But hopefully, you all consider alternatives to ML, comparisons to baselines, how much data you should be training on, and the number of features. And most importantly, what is the bottom line impact of your model translated into real world impacts.
Predicting Police Misconduct: https://www.nber.org/papers/w32432
━━━━━━━━━━━━━━━━━━━━━━━━━
★ Rajistics Social Media »
● Home Page: http://www.rajivshah.com
● LinkedIn: https://www.linkedin.com/in/rajistics/
━━━━━━━━━━━━━━━━━━━━━━━━━
source
