Report System

Name: NextStat
Author: Andrei Toistev

nextstat report generates publication-ready numeric artifacts from a HistFactory workspace. All outputs are versioned JSON schemas; optional Python rendering produces multi-page PDF and per-plot SVGs.

Quick Start

# Generate all artifacts from a workspace
nextstat report --workspace workspace.json --out-dir results/

# Include rendering (requires matplotlib)
nextstat report --workspace workspace.json --out-dir results/ --render

# Blinded mode (masks observed data)
nextstat report --workspace workspace.json --out-dir results/ --blind

Artifacts

File	Description
distributions.json	Prefit/postfit expected yields per sample per region, bin edges, data, Garwood errors, ratio
pulls.json	Nuisance parameter pulls and constraints (postfit sigma / prefit sigma)
corr.json	Correlation matrix from inverse Hessian (optional raw covariance via `--include-covariance`)
yields.json / .csv / .tex	Per-region per-sample prefit/postfit yield tables
uncertainty.json	Ranking-based uncertainty breakdown (skippable via `--skip-uncertainty`)

Visualization Subcommands

nextstat viz distributions --artifact results/distributions.json
nextstat viz pulls        --artifact results/pulls.json
nextstat viz corr         --artifact results/corr.json
nextstat viz ranking      --artifact results/uncertainty.json

Python Rendering

# Render all artifacts to PDF + SVG
python -m nextstat.report render --input-dir results/ --output-dir plots/

# Requires pip install "nextstat[viz]" (matplotlib)

Flags

--blind — mask observed data in distributions and yields (prefit-only mode)
--deterministic — stable JSON key ordering for reproducible diffs
--uncertainty-grouping prefix_1 — group systematics by name prefix in ranking
--render — invoke Python rendering automatically after artifact generation