Re: Pre-PyGotham Roadmap

[Christopher Riederer]

...and sent that to the wrong google groups email address.

On Thu, Jun 16, 2016 at 12:00 PM Christopher Riederer <christophe...@gmail.com> wrote:

Hello Dplythonistas,

Welcome to the first inaugural email of the dplython email list! The purpose of this list is to have more free-flowing discussion about dplython outside of github issues.

We're a month away from PyGotham (https://2016.pygotham.org/) where I'll give a talk about dplython. Before my talk, there are a few key things that I want to get done. These are:

• Speed. There are (at least) two key areas where dplython is way slower than pandas. The first is mutates and summaries on data with lots of groups. The second is the fact that currently dplython copies the entire dataframe on every >>. I'd like to fix both of these.
• DelayModule. dplython should "play nicely" with all other python code. I'd like to make a simple function that can take a module as input, and will apply DelayFunction to anything callable in that module, so that you can use Laters with it.
• Better docs
• Nice tutorial
• (Possibly) Benchmarks. This might be overkill, but I'm considering adding in a benchmark framework, like the one that pandas uses (asv) to track the performance of dplython over time.
• (some) More dplyr features. At the very least, it would be good to make a definitive list somewhere of all the dplyr features and if they've been built in dplython yet.
  

If you have any opinions on these (things that should be de-prioritized, things that should be higher up, etc.) please respond!

Thanks,

Chris

PS Shout out to Dan Robinson, whose Later refactor is great.

[Sujan Rayamajhi]

Chris , how to use np.where inside dplython. 

[Christopher Riederer]

Hi Sujan
Np.where is not part of dplython. You can still use it on a dplyframe the same way you would on a data frame.

Instead of np.where, in dplython you can use "sift".
For example
diamonds >> sift(X.carat > 3)
Will return a data frame where only rows where carat is larger than 3 is returned.

Thanks
Chris

--
You received this message because you are subscribed to the Google Groups "dplython" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dplython+u...@googlegroups.com.
To post to this group, send email to dply...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dplython/53e8e26b-3b8c-451f-b904-0856b52e67d1%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.