How swethamagesh05/anti-burnout-keyboard works
This is a keystroke-based burnout/stress estimation app. In the main workflow, a user types in the React interface; it tracks keystroke timing and live WPM, assembles session features, and posts them to FastAPI, which predicts a stress class using a loaded Random Forest model. The interface displays the result, a percentage visualization, and a recommendation. Separate scripts support synthetic dataset generation, model training, and offline keystroke logging and prediction. The sampled sources show these flows; the dataset and model files are represented as artifacts.
Main entry points: App.jsx Typing and results · stress_model.pkl Stress model · main.py Prediction API · dataset.csv Training dataset
Components 13
Typing interface
- React entry
frontend/src/main.jsx - Typing and results
frontend/src/App.jsx
Live prediction
- Prediction API
backend/app/api/main.py - Stress model
backend/models/stress_model.pkl
Model preparation
- Dataset generator
backend/generate_dataset.py - Training dataset
backend/data/dataset.csv - Model trainer
backend/train_model.py
Offline analysis
- Keystroke logger
backend/keyboard_logger.py - Keystroke records
backend/data/keystrokes.csv - Feature extractor
backend/feature_extractor.py - Session predictor
backend/predict_stree.py - Dataset predictor
backend/predict.py
Other components
- User
Connections 14
- User → Typing and results: types and requests
frontend/src/App.jsx - React entry → Typing and results: renders
frontend/src/main.jsx - Typing and results → Prediction API: posts features
frontend/src/App.jsx - Prediction API → Stress model: loads and predicts
backend/app/api/main.py - Prediction API → Typing and results: returns stress class
frontend/src/App.jsx - Dataset generator → Training dataset: writes
backend/generate_dataset.py - Model trainer → Training dataset: reads
backend/train_model.py - Model trainer → Stress model: saves
backend/train_model.py - Keystroke logger → Keystroke records: records
backend/keyboard_logger.py - Feature extractor → Keystroke records: reads
backend/feature_extractor.py - Session predictor → Keystroke records: reads
backend/predict_stree.py - Session predictor → Stress model: loads and predicts
backend/predict_stree.py - Dataset predictor → Training dataset: reads
backend/predict.py - Dataset predictor → Stress model: loads and predicts
backend/predict.py