Probabilistic nowcasting and four-week forecasting of influenza, COVID-19 and RSV admissions and ED visits. Independently fitted stages share one model interface and connect through sampled reconstructed histories.
- Documentation
- Architecture and reflection
- Workflow: acquire, build, plan, launch and rank
- Model runs: dated forecasting reports; nowcasting section
- Data and interactive staircase
- Setup
uv sync
.venv/bin/python -m tapestry.dataset.build show
.venv/bin/python -m tapestry.experiment.planner status -e b-2-t0planner rank is the canonical analysis path. Reports live in docs/experiments/;
model artifacts live in data/experiments/. The two-stage nowcasting/forecasting
pipeline remains available through task=pipeline; standalone tasks use
task=nowcast or task=forecast. See the workflow for full manager commands.