Machine Learning Consultation for Unstructured Free-Text
Health Net of West Michigan works to improve the health of their community by bridging the healthcare and social services systems. They work alongside clients from an empowerment perspective to provide tools for problem-solving and advocacy to reduce barriers they might be facing.
Project at a Glance:
Team:
Data Scientist
Hours Engaged:
15
Skills & Stack:
Sentiment Analysis, Machine Learning
Problem
Health Net of West Michigan connects healthcare and social services systems through many resources and events, including workforce training. During these training sessions, they ask all participants for feedback through surveys.
This feedback is free-text and difficult to analyze, particularly in understanding if these trainings are effective across participants’ education levels. Health Net sought a data scientist to advise on best-in-class methods to analyze this free-text information.
Solution
Our matched talent researched and identified three solution options weighing cost, upfront development time, tool requirements, maintenance needs, risks and limitations, and overall benefits of each potential solution. The talent recommended a solution that the nonprofit will move forward with implementing.
All potential solutions were best-in-class tools for sentiment analysis that could work for HealthNet, including custom code with BERTopic modeling, Excel AI Add-Ons such as Chat GPT, and out-of-the-box solutions like Levity AI and Repustate.
"I had a great time working with Prometheus and developing my leadership and technical skills to aid me in creating meaningful work for a better and useful cause." - Umair (Data Scientist, University of Calgary)