WEB: Home page content for the new website

[datapythonista]

In #28014 we added a first draft of the new pandas website. It can be seen here:
https://datapythonista.github.io/pandas-web/ (will be at https://dev.pandas.io when #28497 is merged).

Things that are already in the home page, or that could be added:

• Short sentence to summarize what pandas is about
• Links to the main pages / section (feedback on things to add/remove/change welcome)
• List of companies supporting pandas (do we need to formalize the criteria to be there?)
• Info and links to the last version (download, docs...)
• Follow us on twitter
• Wes book
• Links to previous versions
• Project highlights (See the home of Jupyter, Dask, Django or Spark for reference). How do we want it? What exact highlights do we want to list?
• Success stories. We have some in the current website (https://pandas.pydata.org/#what-do-our-users-have-to-say) but in my opinion we could do much better. Like some known faces/companies. I can ask one from the team of the black hole image for example.
• Summary of the ecosystem page. Like the basics, mentioning matplotlib for plotting, numpy for array computing, Jupyter as an interface, Dask for distributed...
• Anything else?

[WillAyd]

Maybe I'm the only one but I actually dislike the carousel. It feels a little gimmicky to me in that it doesn't necessarily portray important information any more effective than a tutorial would, so I'd rather just get rid of it altogether

[datapythonista]

If I go to a website, and the home page doesn't clarify what the product is about well enough, it's quite likely that I just leave it. Rather than going to the tutorials and spending a significant amount of time on understanding a product I still don't know if I need it or not.

I don't like how the carousel looks right now (I count with the designer to help with that). And we don't necessarily need a carousel. But just removing it makes the home page quite useless to people new to pandas in my opinion.

[TomAugspurger]

Maybe we can repurpose this discussion to: "what should be on the homepage", rather than do we like a carousel / what do we want in it (is that OK @datapythonista?)

I think it'd ideally have

1. Easy to find links to common resources (possibly has a header)

• docs
• install / download
• tutorial / examples
• contribute / github
• about / donate / ...

2. A quick (one sentence) summary. Something like "Pandas provides high performance data structures for tabular data"
3. Components / features. Not sure how to phrase this one, but things like "data structures", "groupby", "joins", "IO", .... Each would be a small section describing what pandas provides. Ideally this would be a mix of visuals and text.
4. Other stuff (used at, supported by, etc.)

Some examples I like

• https://jupyter.org
• https://dask.org
• https://www.djangoproject.com
• https://spark.apache.org

[datapythonista]

Surely happy to discuss more about the home page in general. Since nobody made any comment about it in the PR with the proposal, I assumed everybody was happy, and I was moving forward with the details.

I agree with the things you think the home should have. I think it's very aligned with what's in the draft I've got. If you have anything specific that you'd add, remove or change, that would be useful.

I also like the example websites you linked. With a professional designer helping, I guess we can focus more on the content, and let her propose the final visual aspects.

[datapythonista]

I've been thinking about this, and also discussed with NumPy, who are building a website, and I think we should get rid of almost everything we've got in the current home page, and have something more similar to:

• https://jupyter.org (suggested by Tom)
• https://dask.org (suggested by Tom)
• https://julialang.org/ (suggested by Ralph Gommers)

What I'd have is:

• Very quick summary (like the box we have now, but much cooler)
• Features / highlights of pandas (similar to the 3 mentioned links)
• Ecosystem (similar to Julia)
• Main sponsor links (similar to what we and dask have)

Of what we have now, the 3 columns with the links seem redundant with the top navigation, the follow us is in the footer, I'll move the history of versions to a separate page (linked in the navigation Documentation > Older versions), and I moved Wes book to the getting started page. I'll place the non-redundant information of the "Latest version" part somewhere else.

Any objection?

[TomAugspurger]

Sounds good to me.

[jorisvandenbossche]

I'll move the history of versions to a separate page (linked in the navigation Documentation > Older versions)

This could also be handled in the documentation itself (ideally in the new theme we would have a dropdown to select older versions, like readthedocs offers, or like https://docs.python.org/ also does).
(but fully agree this doesn't need to be so prominent on the home page!)

Ecosystem (similar to Julia)

With the idea that this replaces the separate ecosystem page, or just some highlights out of it that further link there?

---

I think it could be nice to have something of "news" on the home page. Eg how https://spark.apache.org/ does it, IMO it is useful to see that there were recent releases.

[datapythonista]

With the idea that this replaces the separate ecosystem page, or just some highlights out of it that further link there?

Highlights, and keep a separate ecosystem page. I was thinking on having some sort of tabs talking about things like:

• Python: Why Python is awesome
• Plotting: Matplotlib, Bokeh, Altair, and how they interact with pandas
• Distributing pandas: Dask and why Spark sucks (ok, not in this words ;)
• Interface: Jupyter, Jupyter lab, Spyder...
• Conda: Easy to install and get all what is needed
• NumPy: Our brother for array computing
• xarray: Our brother for labeled non-tabular/multidimensional data

Not sure the exact items, but give an overview that we're in great company, and then link to the detailed ecosystem page.

[jorisvandenbossche]

OK, that sounds cool! (I would maybe add xarray somewhere, our brother for labeled non-tabular/multidimensional data)