This was the third project I completed for the Metis data science bootcamp. I investigated prison population data using the 2004 Survey of Inmates conducted for the Bureau of Justice Statistics.

I initially set out to classify whether inmates were imprisoned for violent or non-violent crimes and restricted my models to features which dealt mainly with demographics and lifestyle. My investigation picked up on some interesting things in the data, namely the low rate of inmates with high school degrees, much lower than the general population. The focus of my investigation then shifted to this discrepancy in education rates.

You can read more about my findings on my website.

####Technologies used:

• Python
• Classification algorithms (logistic regression, SVM, naive Bayes, etc)
• D3.js