Predicting Non-Medical Opioid Use in Young Adults
- Category: Data Science
- Purpose: UCLA DataFest 2021
- Award: Judges' Choice Award
- Project date: April 2021
- Video Presentation: Scroll to the Bottom of the Page
- Slides: Slides
- Write Up: Document
The Power of a Logistic Regression
Given an enormous medical dataset, my team and I sought to analyze non-medical opioid use in our own demographic, 18-24 year olds. What variables are strong predictors of opioid abuse? Knowing these factors can optimize preventative measures and support groups by guiding them towards populations that may need them the most. Using R, the four strongest predictors we found were (1) having a mental health disorder, (2) gender, and (3) involvement in greek life. We added an external fourth variable, (4) the legality of recreational marijuana in the state, but we did not see a strong relationship.
Pill image courtesy of: Vascual Health Clinics