DOC: need SQL -> pandas tips

[jreback]

of course #4164 makes this syntax easier anyhow

Here's an example of the types of things we need (obviously not all apply here)
http://www.pytables.org/moin/HintsForSQLUsers

we have this in the docs for join/merge, but maybe useful for:

• groupby
• boolean indexing to replace select

e.g.

select * from mytables where A='bar' and B>to_date('20130101')

replaced by:

df[(df.A=='bar') & (df.B>Timestamp('20130101'))

select count(*) from mytables

replaced by

len(df)

etc

[hayd]

Where were you thinking this should go?

Perhaps could add to/rename the R comparison section cc #3980

Although tbh not sure what purpose of that section actually is, perhaps slightly different? But if it was syntax/idiom differences:

pandas for R users
pandas for SQL users

seem like they should be in the same section.

[cpcloud]

actually you should be able to do df["A == 'bar' and B > 20130101"] ftw.

[jreback]

yes....syntax differences is a better title....maybe a How To section?

e.g. say you are from a SQL backgroud, how to do '......' in pandas?

[RomanPekar]

I think I could try to find some SQL popular / tricky examples and compare it to pandas, where I should put these comparisons? I still don't have good pandas solution for some SQL queries, but it'll be interesting to find it

[jreback]

interesting reference

http://johnbeieler.org/blog/2013/06/06/using-sql/

Here is a notebook you guys can use if you are short in time.

Link

[jtratner]

I thought this was a nice intro - ttp://www.gregreda.com/2013/10/26/intro-to-pandas-data-structures/ (by @gjreda)

[gjreda]

Looks interesting - what'd you have in mind, @jtratner, @jreback? I'd be happy to restructure or contribute the tutorial as much as possible.

[jtratner]

I like how your intro gets through key features (like I/O) in a very human way that's easy to follow. That said, I'm not sure there's a good fit for that sort of thing in the docs (10 minutes to pandas is somewhat like that, though less conversational), but the later parts where you take an SQL statement and convert it into what you'd use in pandas could be a nice fit, especially with your explanation. That plus some of the notes from @pichonz could be a nice foundation for a SQL:pandas 'cheat sheet'.

Also, if you have any suggestions for edits to the docs, those are always welcome!

[gjreda]

Sounds good. I'll try to add some stuff over the next week or two.

[jtratner]

Great!