Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
Sterling Paramore
@gnilrets
Any chance this could be a viable means for more real-time communication? Something else maybe?
Victor Shepelev
@zverok
@gnilrets in fact, @v0dro is pretty responsive on email. And, while I'm not a core maintainer, I'm one of GSoC mentors for this project, and I'm always here, so you can speak here, if you want.
Sterling Paramore
@gnilrets
Cool! I was trying to figure out the best way to add two vectors together. They don't all have the same indices, so I'm getting nils in the places where they don't overlap. Trying to figure out if there's an elegant way to get around this,
The best I've found so far is to apply #detach_index to both vectors, concatenate the two dataframes, group-by the index, then sum. But that that seems pretty round-about.
Victor Shepelev
@zverok
Could you please share some sample code? I may be not the most proficient daru user, but that's what we are trying to achieve now -- the "intuitively right" way of things.
Sterling Paramore
@gnilrets
v1 = Daru::Vector.new([1,2], index: ['A', 'B'])
v2 = Daru::Vector.new([1,2,3], index: ['A', 'B','C'])

v1 + v2 # would expect the C index to be 3, but instead it is nil

#workaround
(v1.detach_index.concat v2.detach_index).group_by([:index]).sum[:values]
I guess we probably need a different operation besides + for what I'm trying to do, because of course things would get tricker if I was trying to -, *, or /. But + is special because it's commutative and has 0 as the identity operator.
Sterling Paramore
@gnilrets
Also, before I sign off for the weekend, I just wanted to give a special thanks to @v0dro for the Daru project. I was so close to reluctantly switching to using Python to do data ingestion and ETL. But Daru has made it possible for me to continue using Ruby, which makes every day at work a pleasure!
Lokesh Sharma
@lokeshh
Can we change the behavior of MultiIndex so that MultiIndex.new does what MultiIndex.from_tuples do now
The benefit will be that at many places where we want to create index. We can simply do @index.class.new subset_values regardless of which index class is there.
Also its a good idea for a subclass to follow the behavior of super class I think.
Sameer Deshmukh
@v0dro
@lokeshh bad idea.
Agreed it will be easier to intialize objects internally... but the current interface makes it explicit to users that they're initializing from tuples.
Sameer Deshmukh
@v0dro
@gnilrets can you share your email ID?
I'll add you to the sciruby slack channel. It's quite responsive.
Lokesh Sharma
@lokeshh
ok
Sterling Paramore
@gnilrets
Cool. I have slack open all the time. gnilrets@gmail.com
Shekhar Prasad Rajak
@Shekharrajak

Hello, I am trying to setup daru in my system, but I am getting some errors when I do

gem install bundler
bundle install
bundle exec rake compile
bundle exec rake spec

Let me know how can I setup development version in my system.

Lokesh Sharma
@lokeshh
Read contributing.md file in the repo
Rick Hull
@rickhull
so when is master going to be fixed?
is there any way to revert or steer clear from breakage in master?
Victor Shepelev
@zverok
@rickhull as far as I can understand, there is kind of rush currently trying to do to much things at once. I hate the fact that bad code was merged, so I'll try to do my best tomorrow to have working master tomorrow. As far I can understand, @v0dro is busy with another project, so probably I am the next in charge :(
Rick Hull
@rickhull
ok, no worries :)
sorry if I've been overbearing, just new to the project :)
no rush at all
Victor Shepelev
@zverok
yeah, it is ok, to be honest the current state is kinda doubtful, though we currently prefer to move quickly (sometimes with ugly consequences) rather than stay conservative
Rick Hull
@rickhull
in the meantime, I have some questions about how to use and reason with data frames
i have a specific example in mind, but let me hang on to it for a bit
it seems to me that a data frame represents a set of parameters across a population (or iteration)
maybe we are looking at 0-100 for a set of individuals, or one individual from t=0-100
but one event has a consistent representation whether xy, xyz, or abcwtfbbq
Rick Hull
@rickhull
an event is a set of associated attributes, no?
is that event also a vector? sure
but usually we talk about one attribute over say time, that's a vector
so one view is the xyz is a vector, and we have N of them
the other view is x0..N is a vector
and we have two more
y and z
Rick Hull
@rickhull
in daru, there is a row / vector distinction
how is this made? is it arbitrary?
Rick Hull
@rickhull
let me make this more concrete. say I have 3D coordinates over time. xyz t=0..N
perhaps a table of x, y, z, t
maybe the t intervals are regular, maybe not
which are the vectors and rows if I want to meadure central tendency in 3D space
*measure -- sorry, busted elbow
Athitya Kumar
@athityakumar
@rickhull @zverok : Build is passing now. :smile:
Victor Shepelev
@zverok
:tada: