RubyData Workshops

[Andrei Beliankou]

Dear list members,

meanwhile we are not a small community but we still need to invite new
(active) members. And one of good ways is to show where Ruby can be used
in the real life (also including web dev).

Python devs are used to present their work at a variety of conferences
and workshops.

I'd suggest to start a regular event on a wide range of data related
Ruby applications: ML, DataScience, Math, NLP, CL, Statistics,
Chemistry, Physics etc.

We don't have the ability to meet in person: no money for now, no time,
to location, no sponsors. But we can change this situation if we meet
online, e.g. via a Hangout which will be recorded and put online.
YouTube is a good place to spread our presence and get attention among
other Ruby and non-Ruby users.

It shouldn't not be an event of today, or tomorrow or even of the day
after tomorrow, but let's think about that.

Some questions we have to consider:
1. Is there any event we can join and not create our own?
2. Which format? How often? Which time to include as many people as
possible?
3. How to get attention to this event?

Any feed back is welcome :)

Best,
Andrei

[John Woods]

I think a key problem is finding someone with the time and resources to administer/manage the community. As I'm no longer in academia, my time is much more closely managed.

With that said, yes — all of these are good, important ideas. I've been meaning to publish something on NMatrix for years, but it hasn't gotten done. We probably don't have the resources for conference travel.

--
You received this message because you are subscribed to the Google Groups "SciRuby Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sciruby-dev...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Sameer Deshmukh]

The Japanese community is getting pretty active in this regard. I also speak at whatever conferences I can anywhere in the world (spoken at 5 till now).

One of the chief problems is that the Ruby community coming for Ruby conferences primarily consists of web developers and its hard to sell scientific tools to them. I hope daru-views and daru-io go some way in cementing that.

We should probably come up with something like a 'workshop guide' for people who want to conduct RubyData workshops at Ruby conferences.

Have a look at this DataScience.rb workshop by Kenta Murata: http://tech.speee.jp/entry/2017/05/23/080000

[Kenta Murata]

We have different projects which make different data tools with different policies.

As SciRuby is one of them, I think we need another place that bridges between these projects.

So I'm working for launching RubyData community that is like PyData, but especially for Rubyists.

In RubyData, we want to provide three things:

- Environments to try open source data tools with Ruby

- Documents to learn such tools and related topics

- Places to collaborate with each other

The purpose to organize RubyData is increasing the following two populations:

- Developers who write opensource data tools for Rubyists

- Users who use opensource data tools with their products written in Ruby

Moreover, we want to make borderless world between projects and between other language communities.

In order to make it, we want to make RubyData a bridge between projects, such as SciRuby and Ruby Numo, and to cooperate with other organizations such as PyData.

Currently, RubyData only has landing page, some docker images, and one workshop material.

- http://ruby-data.org/

- https://github.com/RubyData/RubyData.github.io

- https://hub.docker.com/r/rubydata/

- https://github.com/RubyData/docker-stacks

- https://github.com/RubyData/rubykaigi2017

And the first RubyData Tokyo Workshop will be held on 26 Oct.

This workshop consists of the same contents of the workshop held in RubyKaigi 2017.

Any contribution is very welcome.

If you are interested in this concept, please join this community as an organizer.

If you want to hold a workshop in your local city, please do it.

Regards,
Kenta Murata

[Andrei Beliankou]

To put my two cents I think that the distiction between SciRuby and
RubyData as parallel to SciPy and PyData could be really successul if we
see SciRuby as the technological community around the scientific stack
and the RubyData as an info event.

This definitions are very important since otherwise we'll get broader
fragmentation. I'm not aware of any central place of Ruby Data related
news. If we'll create an neutral place for information exchange
everybody will benefit from it.

The question for now is how to join the community? :)

On 10/04/2017 05:00 PM, Kenta Murata wrote:
> We have different projects which make different data tools with
> different policies.
> As SciRuby is one of them, I think we need another place that bridges
> between these projects.
>
> So I'm working for launching RubyData community that is like PyData, but
> especially for Rubyists.
>
> In RubyData, we want to provide three things:
>
> - Environments to try open source data tools with Ruby
> - Documents to learn such tools and related topics
> - Places to collaborate with each other
>
> The purpose to organize RubyData is increasing the following two
> populations:
>
> - Developers who write opensource data tools for Rubyists
> - Users who use opensource data tools with their products written in Ruby
>
> Moreover, we want to make borderless world between projects and between
> other language communities.
> In order to make it, we want to make RubyData a bridge between projects,
> such as SciRuby and Ruby Numo, and to cooperate with other organizations
> such as PyData.
>
> Currently, RubyData only has landing page, some docker images, and one
> workshop material.
>
> - http://ruby-data.org/
> - https://github.com/RubyData/

> <https://github.com/RubyData/docker-stacks>RubyData.github.io

> - https://hub.docker.com/r/rubydata/
> - https://github.com/RubyData/docker-stacks
> - https://github.com/RubyData/

> <https://github.com/RubyData/docker-stacks>rubykaigi2017

>> <javascript:>.

>> For more options, visit https://groups.google.com/d/optout

>> <https://groups.google.com/d/optout>.

>>
>> --
>> You received this message because you are subscribed to the Google
>> Groups "SciRuby Development" group.
>> To unsubscribe from this group and stop receiving emails from it, send
>> an email to sciruby-dev...@googlegroups.com

>> <mailto:sciruby-dev...@googlegroups.com>.

>> For more options, visit https://groups.google.com/d/optout.
>
> --
> You received this message because you are subscribed to the Google
> Groups "SciRuby Development" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sciruby-dev...@googlegroups.com

> <mailto:sciruby-dev...@googlegroups.com>.

[Kenta Murata]

Andrei, I saw your post: http://www.rubyflow.com/p/hk40ll-rubydata-workshops

Thank you very much for spreading RubyData.

By the way, I didn't know this site until I found your post in twitter timeline.

Is RubyFlow more famous announcement place than ruby-talk ML?

> The question for now is how to join the community? :)

Now I use a Trello board to manage tasks and notes to launch RubyData.

You can see it at https://trello.com/b/fXuxcw4d

Andrei, I want you to add cards and comments to note anything you noticed.

Can I add you as a member of this board?

The current biggest problem is that we don't have our official space to collect people.

I'm working to host discourse (*1) platform as Julia community did, and I want to finish this by the next Monday.

(*1) https://github.com/discourse/discourse

I need some help to organize discourse after launching it.

I will migrate our management Trello board into this discourse.

Regards,
Kenta Murata

To unsubscribe from this group and stop receiving emails from it, send an email to sciruby-dev...@googlegroups.com.

[Andrei Beliankou]

Dear @mrkn,

On 10/05/2017 02:54 AM, Kenta Murata wrote:
> Andrei, I saw your post: http://www.rubyflow.com/p/hk40ll-rubydata-workshops
> Thank you very much for spreading RubyData.

As you know I try to aggregate and communicate data related news for
Ruby e.g. via the @nonwebruby Twitter account and on the awesome lists.
The reason is rather easy: Ruby is a language for everything in Japan,
in Europe not really. So we all crete jobs for us in the future.

> Is RubyFlow more famous announcement place than ruby-talk ML?
>

To my knowledge the RubyFlow and for sure RubyWeekly by Peter Cooper
(sourcing many posts from the RubyFlow) are high valued resources.
I get many news from these sources but sure from many other souces on my
RSS list.
All in all we'll need many announcement places:
1. Twitter (using #RubyML, #RubyDataScience, #RubyNLP)
2. Ruby ML
3. RubyFlow
4. Reddit
5. HackerNews
6. SciRuby ML
7. SciRuby Slack Channel
8. RedData Gitter
9. Something else?

>> The question for now is how to join the community? :)
>
> Now I use a Trello board to manage tasks and notes to launch RubyData.
> You can see it at https://trello.com/b/fXuxcw4d
>
> Andrei, I want you to add cards and comments to note anything you noticed.
> Can I add you as a member of this board?

Sure, do that please.

>
> The current biggest problem is that we don't have our official space to
> collect people.
> I'm working to host discourse (*1) platform as Julia community did, and
> I want to finish this by the next Monday.

That's very important since the Ruby Forum is read only...

One of my smaller projects is for now to incorporate all awesome lists
in the new newly reborn RubyToolbox. That will get more attention to the
existing libraries.

Thank you a lot for your engagement!

Best,
Andrei Beliankou

[Pjotr Prins]

On Thu, Oct 05, 2017 at 09:09:23PM +0200, Andrei Beliankou wrote:
> Dear @mrkn,
>
> On 10/05/2017 02:54 AM, Kenta Murata wrote:
> > Andrei, I saw your post: http://www.rubyflow.com/p/hk40ll-rubydata-workshops
> > Thank you very much for spreading RubyData.
>
> As you know I try to aggregate and communicate data related news for
> Ruby e.g. via the @nonwebruby Twitter account and on the awesome lists.
> The reason is rather easy: Ruby is a language for everything in Japan,
> in Europe not really. So we all crete jobs for us in the future.

Great initiatives :). There are quite a few among us who have to use
different languages for their daily work. Even when being able to
choose a language one has to be pragmatic - and I am forced to do much
of my programming in languages I am not naming here (oh, shucks,
Python and C++ ;).

Even so, my love for Ruby is there and Ruby really shines when I am
hacking stuff. I am definitely happier when programming Ruby. The only
time I was really unhappy with Ruby was when writing multithreaded
code. But for that we have Elixir (or D). Don't write multithreaded
code in Ruby (or Python).

Ruby: productive language sums it up for me. Some people have
that with Lisp, and I think Ruby is a Lispy language. Ruby
is a Lisp with a nicer syntax than Lisp. The syntax thing I came to
appreciate quite late. Ruby code is easy to read because it *helps*
you read it. Elixir is even more Lispy than Ruby without OOP and with
its macros. But I must say that every time Elixir bugs me it is
exactly because a macro is involved. But then macros can be powerful
too. So it is beauty gained vs time wasted... Ruby macro's would be
cool, but you can just see where it would go wrong. Ruby proves you
actually don't need much of that stuff.

OK, Ruby has warts, and some short comings, but it is much better than
Python (or R). It is a shame so many (scientists) are buying into
Python. I can't think of anything Python is better at as a language. I
should know. OK, matplotlib at this point - hardly a feature of the
language.

Anyone here willing to work on a Ruby matplotlib with our guidance and
support? It can become a GSoC project. A Ruby matplotlib would be so
much nicer! We need dynamic and interactive graphics to up the ante.

Pj.

[Andrei Beliankou]

> Great initiatives :). There are quite a few among us who have to use
> different languages for their daily work. Even when being able to
> choose a language one has to be pragmatic - and I am forced to do much
> of my programming in languages I am not naming here (oh, shucks,
> Python and C++ ;).

I'm a big fan of diversity. That's why I like the efforts by @mrkn: his
Pycall is the way we have to go (at least for now) and then use more and
more Ruby native tools. It's a wonderful feeling if I can do Python with
R and Ruby on the JVM.
So we should embrace the world and leverage benefits, not encapsulate
ourselves in the Ruby ecosystem.

>
> Anyone here willing to work on a Ruby matplotlib with our guidance and
> support? It can become a GSoC project. A Ruby matplotlib would be so
> much nicer! We need dynamic and interactive graphics to up the ante.

I was dreaming for a long time about implementing the Grammar of
Graphics for Ruby. I hope I could start in December.

[Pjotr Prins]

Hi Andrei,

On Fri, Oct 06, 2017 at 10:56:05AM +0200, Andrei Beliankou wrote:
> I was dreaming for a long time about implementing the Grammar of
> Graphics for Ruby. I hope I could start in December.

would it be possible to draft what you would like the grammar to look
like? Say for a number of common visualisations. Ideally with some
form of interactivity.

That can be done without coding. It would help us develop ideas of
where to go with this.

Pj.

[Andrei Beliankou]

Actually my plan was to port the ideas from:
1. https://github.com/yhat/ggpy
2. https://github.com/has2k1/plotnine

Hadley Wickham's implementation for R could be the reference.

There are two interface problems:
1. R ist not really rubyish so mirroring the ggplot's interface is a bit
ugly.
2. a completely new interface builds a new learning burden for
experienced R developers.

Using positional vs. keyword arguments vs. plain hash arguments is
another question.

A barplot impolementation could look like:
$ inlcude GG
$ ggplot(dataset, aes(x = var1, y = var2)) + geom_barplot

Everything here is not even a draft, so every suggestion is valuable :)

[Kenta Murata]

On Oct 6, 2017 14:40 +0900, Pjotr Prins <pjotr...@gmail.com>, wrote:

Great initiatives :). There are quite a few among us who have to use
different languages for their daily work. Even when being able to
choose a language one has to be pragmatic - and I am forced to do much
of my programming in languages I am not naming here (oh, shucks,
Python and C++ ;). 

Thank you very much.  We share the same motivation: using Ruby for data utilization.

So I think we can cooperate by overcoming the little differences about our development policies.

OK, Ruby has warts, and some short comings, but it is much better than
Python (or R). It is a shame so many (scientists) are buying into
Python. I can't think of anything Python is better at as a language. I
should know. OK, matplotlib at this point - hardly a feature of the
language. 

As I'm working as a Ruby committer, Ruby is better than Python for me.

But I don't think Ruby is *much* better than Python.

The language differences between Ruby and Python are little things for many people.

Rather, what can they do is most important things for them.

Moreover, the reliabilities of the development of tools are the next anxious things.

I think it is the reason why a lot of people gathered around of Python.

It means Ruby need Ruby's unique killer products for data utilization.

I think that such products can easily be used with web stacks like Rails, Sinatra, and Hanami.

And such products shouldn't prohibit their users from using them with other mainstream data tools,

such as Spark, pandas, and Stan.

Anyone here willing to work on a Ruby matplotlib with our guidance and
support? It can become a GSoC project. A Ruby matplotlib would be so
much nicer! We need dynamic and interactive graphics to up the ante. 

Now we're working for launching RubyData discourse service.

After launching it, please utilize there to gather developers and users of Ruby's matplotlib.

Regards,
Kenta Murata

[Kenta Murata]

On Oct 6, 2017 04:09 +0900, Andrei Beliankou <ar...@yandex.ru>, wrote:

To my knowledge the RubyFlow and for sure RubyWeekly by Peter Cooper
(sourcing many posts from the RubyFlow) are high valued resources.
I get many news from these sources but sure from many other souces on my
RSS list.
All in all we'll need many announcement places:
1. Twitter (using #RubyML, #RubyDataScience, #RubyNLP)
2. Ruby ML
3. RubyFlow
4. Reddit
5. HackerNews
6. SciRuby ML
7. SciRuby Slack Channel
8. RedData Gitter
9. Something else? 

This list is very helpful for me.  I'll check them periodically.  Thanks.

Now I use a Trello board to manage tasks and notes to launch RubyData.
You can see it at https://trello.com/b/fXuxcw4d

Andrei, I want you to add cards and comments to note anything you noticed.
Can I add you as a member of this board?

Sure, do that please. 

I've added you to the board.

The current biggest problem is that we don't have our official space to
collect people.
I'm working to host discourse (*1) platform as Julia community did, and
I want to finish this by the next Monday.

That's very important since the Ruby Forum is read only...

One of my smaller projects is for now to incorporate all awesome lists
in the new newly reborn RubyToolbox. That will get more attention to the
existing libraries. 

We encountered the problem to launch discourse, and now we're working for resolving it.

The problem is that discourse's official distribution is not on the standard docker way.

As I try to launch discourse on AWS Elastic Beanstalk, it is not easy due the problem.

But I think we can resolve for some days.

In parallel, I want to discuss the design of the category structure of RubyData discourse.

I created a candidate categories tree by almost copying from Julia's discourse as a role model [1].

```

- category: Announcements

  description: Low traffic category for important announcements, mostly product releases and workshop announcements

- category: Usage

  description: Question and discussion about using data tools for Ruby.  If you are new to data tools for Ruby or have questions regarding your first program please use the First Steps subcategory.

  sub_categories:

    - category: First steps

- category: Domains

  description: Discussion of Ruby in various specialized subject domains: statistics, optimization, machine learning, linear algebra, networking, GPUs, IDEs, etc.

  sub_category:

    - category: Statistics

    - category: Numerics

    - category: GPU

    - category: Bio

    - category: Data

    - category: Visualization

    - category: Geo

    - category: Optimization

    - category: Machine Learning

    - category: Modeling & Simulation

    - category: Signal Processing

    - category: Web Stack

    - category: Astro/Space

    - category: Finance/Economics

    - category: Parallel/Distributed

- category: Community

  description: Non-technical topics of the RubyData communities, including announcements, founding/deprecation of RubyData related websites or local communities, new publications or other resources, advertisements, etc.

  sub_category:

    - category: Jobs

    - category: Meetups

- category: Meta discussion

  description: Meta-discussion about this site, its organization, how it works, and how we can improve it.  Feedback for other parts of *.ruby-data.org is welcome as well.

- category: Offtopic

  description: Topics that are not about RubyData or do not fit in any other category

```

How do you think this categories?

Are there enough categories for SciRuby?

How do you think about these categories, Andrei?

I want to your opinion as the maintainer of awesome lists.

[1] https://discourse.julialang.org/

Regards,
Kenta Murata

[Alexej Gossmann]

My 2 cents :)

> I was dreaming for a long time about implementing the Grammar of Graphics for Ruby. I hope I could start in December.

I've been thinking about it a lot too. What do you think would be the use cases when someone may prefer using a Ruby grammar of graphics implementation over R's ggplot2?

I use ggplot2 in R heavily. The way I see it, the essential selling points of ggplot2 are (1) the intuitive interface of ggplot2 (after learning the "grammar" I almost never have to look at the documentation) combined with (2) the inherent interactivity of R, as well as (3) the amount of fine-grain control and options you have (e.g., facets, math symbols in plot titles and labels, different smoothing functions and stats build in, themes, extensions, etc.). 

I think in Ruby one could imitate (1), or even improve on it because Ruby is more powerful than R. As for (2), I don't think Ruby can reach it: when doing exploratory data analyses in Ruby I find the loss of convenience very noticeable (Ruby is more of a conventional scripting language, unlike R, which is an interactive interface for data analysis). Point (3) would just require very serious amounts of time. In addition another great feature is how smooth ggplot2 works with other R packages (such as dplyr, tidyr, lubridate, etc.). What do you think?

About Python, I think that currently one of the greatest attractions of Python in data science and machine learning is all things deep learning (other than deep learning Python does not offer anything for me that is not covered by an intersection of Ruby and R). All the deep learning things which are imho worth learning currently are Python-first (Tensorflow, PyTorch, Keras, etc.). I would really love to have a Ruby Keras. Let me know if you have any ideas on that. There is even an R interface, but I don't know how it is done.

Alexej

--
You received this message because you are subscribed to the Google Groups "SciRuby Development" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sciruby-dev+unsubscribe@googlegroups.com.