Datetime formatting in pandas .to_markdown()

user_11021886 0
Viewed 355

I have a pandas DataFrame which has a column of dtype datetime:

import pandas as pd

# Mock-up data
df = pd.DataFrame({'year': [2015, 2016],
                   'month': [2, 3],
                   'day': [4, 5]})
df = pd.to_datetime(df)

print(df)

# 0   2015-02-04
# 1   2016-03-05
# dtype: datetime64[ns]

I would like to use the .to_markdown() method to display this DataFrame.

However, the .to_markdown() method displays the datetimes in scientific notation:

print(df.to_markdown())

# |    |           0 |
# |---:|------------:|
# |  0 | 1.42301e+18 |
# |  1 | 1.45714e+18 |

Is there a way to have the .to_markdown() method display these dates in a more human-readable manner? The .to_latex(), .to_csv(), and .to_string() methods already behave this way:

# Other .to_ methods behave as desired, eg.
print(df.to_latex())

# \begin{tabular}{ll}
# \toprule
# {} &          0 \\
# \midrule
# 0 & 2015-02-04 \\
# 1 & 2016-03-05 \\
# \bottomrule
# \end{tabular}

pandas version: 1.3.2

tabulate version: 0.8.9

pandas markdown python
1 Answers

Under the hood the .to_markdown() method uses the tabulate package. The floatfmt named argument can be used to control the formatting of floats, but I cannot see how this could be useful here.

The best solution I can currently find is simply to format the datetime column as a column of strings before calling the .to_markdown() method:

print(df.astype(str).to_markdown())

# |    | 0          |
# |---:|:-----------|
# |  0 | 2015-02-04 |
# |  1 | 2016-03-05 |
user_110218860
Related