`pandas.to_latex` - how to make column names bold

Question

When I'm using the pandas.to_latex function to create latex table, the column names are unfortunately not bold. What can I do to make it bold?

Answer 1

Update

I have been told on GitHub that this is allready possible with plain pandas but there is some missing documentation, which will be updated soon.

You can use the line below.

result = df.style.applymap_index(
    lambda v: "font-weight: bold;", axis="columns"
).to_latex(convert_css=True)

Old answer

Here is a complete example, this is adapted from the offical documentation.

There is a keyword to print bold columns bold_rows=True. Sadly there isn't a kyword parameter to do the same for the columns. But I can use this to check if my code gives the same result for the column headers.

I use the result of to_latex() and split it in three sections. One section is the line with the column names. In this line I use a regular expression to add the \text{}-string. My code works only if you colum names don't have a whitespace.

import pandas as pd

df = pd.DataFrame(
    dict(name=['Raphael', 'Donatello'],
         mask=['red', 'purple'],
         weapon=['sai', 'bo staff']
        )
)
ans = df.to_latex(bold_rows=True)

split_middle = '\n\midrule\n'
split_top = '\\toprule\n'
top, mid = ans.split(split_middle)
start, columns = top.split(split_top)

columns = re.sub('(\w)+', '\\\\textbf{\g<0>}', columns)

result = split_middle.join([split_top.join([start, columns]), mid])
>>> result
\begin{tabular}{llll}
\toprule
{} &       \textbf{name} &    \textbf{mask} &    \textbf{weapon} \\
\midrule
\textbf{0} &    Raphael &     red &       sai \\
\textbf{1} &  Donatello &  purple &  bo staff \\
\bottomrule
\end{tabular}

In the output you can see, that the header now is bold.

Answer 2

Based on mosc9575's intuition to manipulate output of df.to_latex, you can make column header is wrapped with some unique delimiters to make the substitution easier.

import re

s = re.sub('<([^>]*)>', '\\\\textbf{\g<1>}',
           df.rename(columns=lambda x: f'<{x}>').to_latex(index=False))

\begin{tabular}{lll}
\toprule
   \textbf{name} & \textbf{mask} & \textbf{weapon} \\
\midrule
  Raphael &    red &      sai \\
Donatello & purple & bo staff \\
\bottomrule
\end{tabular}

Answer 3

The answers given here are somewhat unnecessary. You can use the pandas.Styler implementation.

styler = df.style
styler.applymap_index(lambda v: "font-weight: bold;", axis="index")
styler.applymap_index(lambda v: "font-weight: bold;", axis="columns")
styler.to_latex(convert_css=True)

The documentation for Styler.to_latex gives alternative ways if you want to customise the LaTeX bold function, e.g. using \textbf{} instead of \bfseries for example.

The Styler implementation is ultimatelty intended to replace the older and less flexible DataFrame.LatexFormatter