Auto-Generated Tables

Never Type a Results Table by Hand

Hand-typed result tables are error-prone and hard to update. When your simulation produces new results, the table should regenerate automatically. Pandas to_latex() and tabulate make this effortless.

Definition:

pandas.to_latex()

Convert a DataFrame to a LaTeX tabular environment:

import pandas as pd

df = pd.DataFrame({
    'Modulation': ['BPSK', 'QPSK', '16-QAM'],
    'BER': [1.2e-5, 3.4e-4, 2.1e-3],
    'Throughput': [1.0, 2.0, 4.0],
})

latex = df.to_latex(
    index=False,
    float_format='%.2e',
    caption='Simulation Results at SNR = 10 dB',
    label='tab:sim_results',
    column_format='lcc',
)
print(latex)

Key parameters: index, float_format, caption, label, column_format, escape, multirow, multicolumn.

Definition:

tabulate — Versatile Table Formatting

The tabulate library formats data as LaTeX, Markdown, HTML, or plain text:

from tabulate import tabulate

data = [['BPSK', 1.2e-5, 1.0],
        ['QPSK', 3.4e-4, 2.0]]
headers = ['Modulation', 'BER', 'Throughput']

print(tabulate(data, headers, tablefmt='latex_booktabs',
               floatfmt='.2e'))

Use tablefmt='latex_booktabs' for professional tables with \toprule, \midrule, \bottomrule from the booktabs package.

Definition:

Pandas Styler for HTML and LaTeX

df.style provides conditional formatting:

styled = df.style \
    .highlight_min(subset=['BER'], color='lightgreen') \
    .format({'BER': '{:.2e}', 'Throughput': '{:.1f}'}) \
    .to_latex(caption='Results', label='tab:results')

Theorem: Reproducible Table Generation Pipeline

A reproducible reporting pipeline should satisfy:

  1. Source of truth: Raw data in CSV/HDF5 — never in the LaTeX file.
  2. Single command: python generate_tables.py produces all .tex tables.
  3. Version control: Both data and generation scripts are in git.
  4. Idempotent: Running the script twice produces identical output.

This ensures any collaborator can regenerate the paper's tables from scratch.

If you cannot regenerate every number in your paper from a single script, your results are not reproducible.

Example: Auto-Generated Simulation Results Table

Generate a LaTeX table from BER simulation results stored in a CSV file.

Solution
Implementation
import pandas as pd

# Load simulation results
df = pd.read_csv('ber_results.csv')

# Format for publication
table = df.pivot_table(
    values='ber', index='modulation', columns='snr_db'
)
table = table.applymap(lambda x: f'{x:.2e}')

latex = table.to_latex(
    caption='BER Results for Various Modulations and SNR Values',
    label='tab:ber_results',
    column_format='l' + 'c' * len(table.columns),
)

with open('tables/ber_results.tex', 'w') as f:
    f.write(latex)
In LaTeX
\input{tables/ber_results.tex}

Example: Professional Table with booktabs

Create a publication-quality table using booktabs rules.

Solution
Implementation
from tabulate import tabulate

data = [
    ['BPSK',   '1', r'Q(2Eb/N0)Q(\sqrt{2E_b/N_0})',    '1.2e-5'],
    ['QPSK',   '2', r'Q(2Eb/N0)Q(\sqrt{2E_b/N_0})',    '3.4e-4'],
    ['16-QAM', '4', r'32Q(2Eb5N0)\frac{3}{2}Q(\sqrt{\frac{2E_b}{5N_0}})', '2.1e-3'],
]
headers = ['Modulation', 'bits/sym', 'BER Expression', 'BER @10dB']

latex = tabulate(data, headers, tablefmt='latex_booktabs')
print(latex)

Table Format Comparison

Compare how the same data looks formatted as LaTeX, Markdown, and HTML tables.

Parameters

Table Generation Tools

ToolInputOutput FormatsBest For
pandas.to_latex()DataFrameLaTeXData from simulation/CSV
tabulateList of listsLaTeX, Markdown, HTML, textQuick formatting
df.style.to_latex()DataFrameLaTeX with conditional formattingHighlighted results
jinja2 templatesAny dataAny text formatComplex report generation

Common Mistake: Special Characters in to_latex()

Mistake:

LaTeX special characters (_, %, $, &) in DataFrame values cause compilation errors.

Correction:

Use escape=True (default) in to_latex(), or pre-process values to wrap them in ...... for math mode.

Quick Check

What does the booktabs LaTeX package provide?

Colored table cells

Professional horizontal rules (toprule, midrule, bottomrule) without vertical lines

Automatic table numbering

Multi-page table support

Correction:
Professional horizontal rules (toprule, midrule, bottomrule) without vertical lines

booktabs provides clean, publication-quality horizontal rules that replace crude \hline commands.

booktabs

A LaTeX package providing professional-quality horizontal rules for tables, using toprule, midrule, and bottomrule instead of hline.

tabulate

A Python library for formatting tabular data as plain text, LaTeX, Markdown, HTML, and other formats.

Matplotlib + LaTeXProgrammatic Figure Generation