Predicting Risk for Depression via Tweets
Human-Centered Machine Learning Course
Fall 2020
Skills
HCML
D3
Critiquing ML Systems
Workshop Paper Writing
Combining Human Interaction with Machine Learning
In Human-Centered Machine Learning, taught by Stevie Chancellor, my team and I designed and developed an interactive system, seen below, that predicts a user’s risk for depression based on their Twitter data and visualizes the results using a classifier model. After reading and critiquing machine learning systems and papers over the course of the quarter, we were able to apply the principles of human-centered machine learning to our project.
Read our workshop paper draft.
Promoting responsible machine learning for social good by providing mental health support and disclaimers
Making our ML model understandable using transparency and interactivity
The interactive area and pie chart allow users to understand what decisions our model made and WHY it made those decisions.
Explore the model and interface more.