ANN: xarray v0.10.2 release

[Stephan Hoyer]

I'm pleased to announce the v0.10.2 release for xarray.

Highlights include:

- xarray.dot() for summation/products like numpy.einsum().

- Support for writing xarray datasets to netCDF with dask-distributed.

- rolling().construct() method for loading rolling window objects into arrays.

- NumPy ufuncs are now supported on xarray.Dataset objects.

For full details, please see the release notes:

http://xarray.pydata.org/en/stable/whats-new.html

Thanks to everyone who contributed to this release:

Deepak Cherian

Keisuke Fujii

Joe Hamman

Stephan Hoyer

Zac Hatfield-Dodds

Best,

Stephan

[Boris D.]

Hello :)

I didn't knew where to post this, so let's do this here:

I'm willing to use xarray to complete some tasks on my PhD mission, however I'm having a hard time finding example &/or documentation for xarray.

Hopefully there is some posts which helped a bit :

https://rabernat.github.io/research_computing/xarray.html

http://danielrothenberg.com/gcpy/tutorials/xarray_overview.html

The online documentation is good but lacks concrete examples. The concepts aren't easy to assimilate to me / a new python user.
It lacks some graphical example which would help me assimilate much faster than just reading again and again the input / output code.
I know that most Python's docs are built that way, however for operations on 3D arrays it's a bit mind-boggling.

I'm very frustrated to spend so much time searching for explanations / solutions as the concept of what I want to do is so basic

It's just my 2cts and I hope I will be able to become a power user on the long term, and contribute on this since I think It's so essential.

[Stephan Hoyer]

Hi Boris,

Those are some lovely tutorials you link to! We should at least add a link to them from the xarray docs so they are easier to find. Ideally, we would links to "Getting started" tutorials for each type of new xarray user (e.g., pandas users, netCDF users, new Python programmers, etc.).

We have a handful of examples in the online docs already, but certainly not enough (see http://xarray.pydata.org/en/stable/examples.html).

It would be helpful if you could elaborate on which topics in particular had the least helpful explanations, were the hardest to find information about, or could really use a picture. You have a valuable perspective on this as a new user -- it's often hard for those of us who are experts to put ourselves into the novice frame of mind.

Best,

Stephan

--
You received this message because you are subscribed to the Google Groups "xarray" group.
To unsubscribe from this group and stop receiving emails from it, send an email to xarray+un...@googlegroups.com.
To post to this group, send email to xar...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/xarray/9086da47-cbd6-4344-b617-80f422058de2%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ryan Abernathey]

As one of the authors of those tutorials, I agree with Boris' characterization of the xarray documentation.

I think what we need is a new top-level heading in the xarray documentation called "tutorial" which is distinct from both "examples" and "user guide." The tutorial would explain xarray from the ground up with a pedagogical tone.

There are many good tutorials out there, and Joe is leading the development of more for scipy. We should endeavor to integrate these into the xarray docs as an "offical xarray tutorial."

-Ryan

On Tue, Mar 27, 2018 at 3:56 AM, Boris D. <dessimo...@gmail.com> wrote:

Hello,

Of courses I checked the tutorials, indeed it wasn't enough for me.

For example I haven't found yet how to ruen make a dimension coordinate to a non-dimension coordinate.

I'm having a hard time figuring out what's the best way to build efficiently my xarray model.

I'm seeing so much possibilities and some that seems "ok" to me are probably not for reasons I ignore yet.

Here's an example close to the data I'm working with :

10 mobile acquisition cards

One measurement around every 60s.

On each sensor I have :

• Date (Indexed, I resampled my data per minuture because I wanted the axis to match and not to oversample, I hope it's a good approach)
  
• General info: Temperature / Humidity / Pressure
  
• GPS information :  Lat / Long / Alt
  
• Six other measurements (P1/P2 ... P6)
  

Right now I have all those data stored in one dataframe for each acquisition card.

I would like to combines all those dataframes in one xarray to the smartest possible way.

I want to of course plot data on a map, but on the long term to goal is to be able to :

• Research
  
• Real-time plotting
  
• Machine learning for events prediction.
  

So xarray seemed a good idea.

I wrote the syntax of the models I built. And I know it's not the best way.

A model where each dataframe corresponds to an object in data variables : 

______________________________________________

Dimensions:   (Date: 200000, variables: 12)

 

Coordinates:

  * Date  (Date) 

  * Variables   (Variables) 'Altitude' 'Longitude' 'Altitude' 'P1' 'P2' 'P3' ...

 

Data variables:

    Sensor1     (Date, Variables)

    Sensor2     (Date, Variables) 

    Sensor3     (Date, Variables)

______________________________________________

A model where each Data variable is... a variable, therefore we just have 1D for each variable

______________________________________________ 

Dimensions:      (Date: 200000, GPS)

Coordinates:

     SensorName   <U5 'AIT01'

     Long

   * Date         (Date) 

Data variables:

    Temperature  (Date) 

    Humidity     (Date)

    Pressure     (Date)

    PM25         (Date) 

    PM10         (Date)

    PM1          (Date)

    Position (Date, GPS)

______________________________________________

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

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

To post to this group, send email to xar...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/xarray/38e8292f-a172-4fcc-bbe6-081c7b559510%40googlegroups.com.