ENH: Display DataFrame column dtypes (data types) for each variable name when printing

[coatless]

Is your feature request related to a problem?

I wish I could incorporate the data type of each column into the output of the Pandas dataframe when using any print method (with a preference for .to_html()). The feature could contain:

• show_dtypes bool, default False
  • Display DataFrame column dtypes (data types) for each variable name.

Describe the solution you'd like

The current way of displaying Pandas data with HTML gives:

The goal here would be to add a subheader row that incorporates the data type of each column variable from df.dtypes output, e.g.

API breaking implications

If the feature is turned on by default, then any unit tests that assume a specific output feature would break as a new header row would be required.

Describe alternatives you've considered

After the HTML object was saved as a string, I used an adhoc method to add an additional subheader with beautiful soup. The method is a bit cumbersome compared to adding a slight modification into pandas/io/formats/html.py

Additional context

To re-create the screenshot above, please run:

import random
import pandas as pd
import numpy as np
from IPython.core.display import HTML
from bs4 import BeautifulSoup

# Simulate data
mu_o = random.choice([19.7,19.8,19.9,20,20.1,20.2])
std_gen = random.choice([1.3, 1.4, 1.5])
sample_size = random.randint(8, 11)
sample_data =  np.round(np.random.normal(mu_o, std_gen, sample_size), 1)

# Convert data set to pandas
df = pd.DataFrame(data = {
    "id": list(range(1, sample_size + 1)),
    "type": random.choices(["good","okay","bad", "horrible"], k = sample_size),
    "vol": sample_data})

# Name columns
df.columns = ['Subject', 'Type', 'Volume (oz)']

# Adhoc way of creating the desired subheader data
html_subheader_dtype = []
for value in df.dtypes.to_dict().values():
  html_subheader_dtype.append(f'<th> <span style="color:grey;text-align:left;font-size:.75em"> ({value}) </span> </th>')

# Convert to HTML
html_df = df.to_html(index = False, justify = "left")

# Use BeautifulSoup 4 to navigate dom
soup = BeautifulSoup(html_df, "html.parser")

# Create new tag for a table row
subheader_dtype_row = soup.new_tag("tr")

# Format with beautiful soup
row_dtype_data = BeautifulSoup("".join(str(x) for x in html_subheader_dtype), "html.parser")

# Add new BeautifulSoup object into the subheader element
subheader_dtype_row.contents.append(row_dtype_data)
  
# Adding the subheader into the table object
soup.table.thead.append(subheader_dtype_row)

# Convert from bs4.BeautifulSoup to str
modified_df_html = str(soup)

# Display Table
HTML(modified_df_html)

[attack68]

upcoming 1.5.0 has a feature similar to this in Styler.

dev docs 1.5.0 Styler.concat

>>> df = pd.DataFrame(data = {"id":[1,2], "type": ["big", "small"]})
>>> df.style.concat(df.agg([lambda s: s.dtype]).style)
     id    type
0     1     big
1     2   small
  int64  object