Alchohol Risk Detection Dashboard

Sensor-driven heavy-drinking risk detection using smartphone accelerometer time-series and TAC-based labels (UCI Bar Crawl dataset), with an interactive analytics dashboard.

Project Scope

• Sliding-window signal processing pipeline for accelerometer streams.
• Classical ML benchmark (RandomForest / ExtraTrees / LogisticRegression / HistGB / XGBoost).
• Sequence-model benchmark (1D CNN, PyTorch) with subject-aware split.
• Single-page web dashboard for risk trends, participant comparison, and downloads.

Dataset

• UCI Bar Crawl: Detecting Heavy Drinking
• Expected archive path: bar+crawl+detecting+heavy+drinking.zip

Environment

• Python: 3.9 (recommended for this setup)
• Install dependencies:

pip install -r requirements.txt

For sequence model training (1D CNN), install the extra dependency:

pip install -r requirements-sequence.txt

Run End-to-End

python src/prepare_data.py
python src/train.py
python src/evaluate.py

Sequence Benchmark

python src/train_sequence.py --max-windows-per-pid 500

Launch Dashboard

python -m streamlit run src/dashboard.py --server.address 127.0.0.1 --server.port 8501 --server.headless true

Open: http://127.0.0.1:8501

Deployment

Streamlit Community Cloud

1. Push this project to GitHub.
2. Create a new Streamlit app.
3. Set entry file to app.py.
4. Deploy with default requirements.txt.

Render (Web Service)

Use included Procfile:

web: streamlit run app.py --server.port=$PORT --server.address=0.0.0.0

Current Results Snapshot

Classical holdout (reports/test_metrics.json)

Metric	Value
F1	0.6686
Precision	0.5975
Recall	0.7588
ROC-AUC	0.6950

Sequence vs Baseline (reports/model_comparison.json, --max-windows-per-pid 500)

Model	F1	Precision	Recall	ROC-AUC
Baseline LogReg Features	0.4925	0.3530	0.8142	0.5448
1D CNN Sequence	0.5179	0.3706	0.8596	0.5298

Repo Structure

alcohol-detection/
  data/
    raw/
    processed/
  models/
  reports/
  src/
    prepare_data.py
    train.py
    evaluate.py
    train_sequence.py
    dashboard.py
  requirements.txt
  README.md