Why, you may ask?
I’m tired of reviewing projects where people make wildly biased/bigoted assumptions about BMI. Despite the fact that data scientists and analysts are supposed to be objective, many of the notebooks presented here are not (that’s a whole other post [objectivity & subjectivity] on its own for another day). I’ve been meaning to make this suggestion for years now…and now I have the time. (Aren’t you lucky?) 
BMI is a bs (sorry for swearing) number that is not an accurate measure of one’s overall health. It also contributes to weight discrimination and stigma in healthcare, not to mention eating disorders in the general population. BMI doesn’t take into consideration an individual’s age, gender, race, muscle & bone mass, body composition, genetics, or medical history. US health insurance companies use this inaccurate number to charge people more in insurance premiums. Just google ‘bmi isn’t an accurate measure of one’s health’. About a zillion articles will pop up. Or, ask any reputable RDN or CDN. I could put you in touch with one or two. 
So, what’s a different data set you could use? If you want to stick with health insurance data, use ACS data regarding income and health insurance coverage by state, the number of uninsured by state, or insurance costs & median income by zip code are all a good start. Even something like car insurance rates by zip code would be interesting. Anything else, please.
Thank you for listening.
