Equa is a comprehensive system for burnout detection that uses machine learning and a large language model (LLM) to provide a proactive approach to workplace wellness. By analyzing key employee metrics, Equa predicts an individual's burnout risk and offers personalized, AI-driven coaching to support mental well-being.
The project combines robust predictive modeling with empathetic AI guidance to offer a scalable and accessible tool for early burnout detection and intervention.
- Predictive Modeling: Utilizes a Gradient Boosting Regressor model to accurately predict an individual's burnout score based on professional and personal metrics.
- Risk Categorization: Classifies individuals into Low, Moderate, and High-risk categories to provide a clear understanding of their current well-being.
- AI-Powered Coaching: Integrates with the Mistral-7B large language model to deliver personalized, empathetic advice based on the user's predicted burnout risk.
- Interactive Conversation: The AI coach supports natural, conversational follow-up questions, allowing for a more dynamic and helpful user experience.
This project was developed using the "Are Your Employees Burning Out?" dataset, which is sourced from a 2008 survey. The dataset is publicly available through the Harvard Dataverse.
The system's core is a Gradient Boosting Regressor model trained on a real-world dataset of employee metrics. This model demonstrated high predictive accuracy, achieving an R-squared (Mental Fatigue Score is the most significant predictor of burnout.
The predicted score is then passed to an AI wellness coach powered by the Mistral-7B model via the OpenRouter API. This coach generates a personalized plan and interacts with the user. The entire system is deployed as a user-friendly web application using Streamlit.