Add support for Pandas Profiling module

Feature description

We’re like to be able to do this:

import streamlit as st
import pandas as pd
import pandas_profiling as pf

DATE_COLUMN = 'date/time'
DATA_URL = ('https://s3-us-west-2.amazonaws.com/'
            'streamlit-demo-data/uber-raw-data-sep14.csv.gz')

@st.cache
def load_data(nrows):
    data = pd.read_csv(DATA_URL, nrows=nrows)
    def lowercase(x): return str(x).lower()
    data.rename(lowercase, axis='columns', inplace=True)
    data[DATE_COLUMN] = pd.to_datetime(data[DATE_COLUMN])
    return data

data = load_data(100)

# Get profile report 
report = pf.ProfileReport(data)

# 👇Show report in Streamlit app! (This is what we want to implement)
st.write(report)

…which should give you something like this:

More info on pandas_profiling here: https://towardsdatascience.com/a-better-eda-with-pandas-profiling-e842a00e1136

Implementation details

The report object in the script above actually has a report.html attribute that we’d be tempted to use for this. However, that’s a bad idea! This would mean we’d be sending raw HTML over the wire and loading it unsafely via innerHTML.

So before implementing this we need to do a little research…

One possibility: serve it in an iframe. See #686 for related work.

Related discussions

https://discuss.streamlit.io/t/including-pandas-profiling-report-in-streamlit/473/4

[sbrugman]

Pandas-Profiling author here. Pandas-Profiling supports multiple user interfaces (HTML report, CLI, Jupyter Notebook/Lab Widgets and PyQt application).

I’d like to invite interested users/developer to contribute a streamlit interface. It can be based on the Jupyter interface and shouldn’t be much work. The relevant code can be found here.

For first-time contributors: don’t worry about it not being perfect at first. We will do a code review from the Pull Request and take it from there.

[MarcSkovMadsen]

I just created an example in Panel using the Pandas Profiling .to_html method and an iframe.

I think you can find the necessary inspiration there.

See

• awesome-panel.org
• Discourse question
• Code