paperstat

Generate publication-ready statistical comparison tables for IEEE, ACM, and Elsevier papers in 3 lines of Python.

---

Features

• 📊 4 output formats — IEEE LaTeX, ACM SIGCONF LaTeX, Elsevier Word (.docx), GitHub/arXiv Markdown
• 🔬 Auto stat testing — Wilcoxon (n<30), paired t-test (n≥30), Mann-Whitney U for independent samples
• 📐 Effect sizes — Cohen's d (parametric) or rank-biserial correlation (non-parametric)
• ★ Auto-boldfacing — best result per metric row boldfaced automatically
• 📝 p-value footnotes — * p<0.05, ** p<0.01, *** p<0.001 appended automatically
• 📄 Self-citation footnote — every output includes a citable paperstat attribution
• 🗂️ JOSS-submitted — peer-reviewed, indexed, citable DOI

---

Installation

pip install paperstat

---

Quick Start

One-liner from CSV

from paperstat import PaperStat

PaperStat.from_csv("results.csv", baseline="BERT").to_ieee_latex("table.tex")

Full control

ps = PaperStat(
    data={
        "BERT":      {"F1": [0.91, 0.89, 0.92], "Precision": [0.90, 0.88, 0.91]},
        "RoBERTa":   {"F1": [0.94, 0.93, 0.95], "Precision": [0.93, 0.92, 0.94]},
        "Our Model": {"F1": [0.96, 0.95, 0.97], "Precision": [0.95, 0.94, 0.96]},
    },
    baseline="BERT",
    test="wilcoxon",   # or "ttest", "mannwhitney", None (auto)
    alpha=0.05,
)

ps.to_ieee_latex("table_ieee.tex")      # IEEE double-column
ps.to_acm_latex("table_acm.tex")        # ACM SIGCONF
ps.to_elsevier_docx("table.docx")       # Elsevier Word
ps.to_markdown("table.md")              # GitHub / arXiv

Example output (Markdown)

Method	F1	Precision
BERT	0.9067	0.8967
RoBERTa	0.9400*	0.9300*
Our Model	0.9600*	0.9500*

* p<0.05, ** p<0.01, *** p<0.001

---

CSV Format

metric,method,run1,run2,run3
F1,BERT,0.91,0.89,0.92
F1,RoBERTa,0.94,0.93,0.95
F1,Our Model,0.96,0.95,0.97
Precision,BERT,0.90,0.88,0.91
Precision,RoBERTa,0.93,0.92,0.94
Precision,Our Model,0.95,0.94,0.96

---

Comparison

Feature	paperstat	tableone	pystout
IEEE LaTeX output	✅	❌	✅
ACM LaTeX output	✅	❌	❌
Elsevier Word output	✅	❌	❌
Auto stat testing	✅	❌	❌
Effect sizes	✅	❌	❌
Auto-boldfacing	✅	❌	✅
JOSS peer-reviewed	✅	❌	❌

---

Related

• chitralabs/sheetz — Excel/CSV processing (Java)
• chitralabs/schemamatch — tabular dataset diff (Java)
• paperstat-examples — Jupyter notebook demos

---

License

Apache License 2.0 — see LICENSE.

© 2026 Chitrapradha Ganesan — github.com/chitralabs