U-SQL Advanced Analytics: Introducing Python Extensions for U-SQL

  • [アーティクル]

Last week at Microsoft's Connect 2016 conference, we announced the General Availability of Azure Data Lake Analytics. As part of the announcement we revealed that U-SQL now includes built-in support for Advanced Analytics scenarios. This includes:

  • The ability to perform massively distributed analytics using Python
  • The ability to perform massively distributed analytics using R
  • Built-in Cognitive capabilities (such as image object detection, sentiment analysis, etc.)

In this post we'll give a very brief overview of the Python support. We'll publish additional blog posts that cover R and the Cognitive scenarios later this week. Below is a very simple "Hello World" using Python that illustrates how easy we've made it to use Python with U-SQL. This is the simplest script that demonstrates how you can run Python on vertexes using a special built-in Python Reducer.  This script shows the key steps:

  1. using REFERENCE ASSEMBLY to bring in the needed Python support
  2. using REDUCE to partition the input data on a key
  3. a built-in reducer (Extension.Python.Reducer) that runs Python code on each vertex assigned to the reducer
  4. Embedded Python code in the U-SQL script that accepts a pandas DataFrame as input and returns a pandas DataFrame as output.

 

To see an simple Hello World sample, go here: https://github.com/Azure-Samples/usql-python-helloworld

To learn more about our support for U-SQL Advanced Analytics and how to enable it in your Data Lake Analytics Accounts, see our Getting Started guide .

Comments

  • Anonymous
    December 02, 2016
    Hi,how do use numpy in this context in terms of loading modules?Do i have to load numpy as an import statement?
    • Anonymous
      December 02, 2016
      Yes, your script has to explicitly "import numpy"
      • Anonymous
        December 08, 2016
        I guess I would be correct in assuming that "gensim" and "nltk" would have to be similarly imported.
  • Anonymous
    December 06, 2016
    Hi, I only get the messages "Assembly master.ExtPython" does not exist.regards,Uli
    • Anonymous
      December 06, 2016
      Hi, just found that python assemblies are copied as part of U-SQL extensions. See the "getting started guide" above.Uli
  • Anonymous
    October 17, 2017
    The "Getting Started Guide" seems to no longer be available as the OneDrive for Business link has expired. Can someone repost this guide?