Difficulty in Installing GraphLab for python 3.5.2 - python-3.x

I am a newbie in Machine learning and want to build a recommendation system for my class project. To do so, I need to work with GraphLab. But I cannot find any version of it which will be supported by python 3.5.2. The latest available version is 2.1, which is supported by python 2.7x. Is there any equivalent package of GraphLab, or any alternative ways to build this system?

Currently there's no stable release of GraphLab for python 3.5. They are working on a stable release. However, there were a beta release here for 3.5 which seems to be inaccessible right now. You may try out scikit-learn, H2O as alternative.They have support for 3.5.

Related

How to determine which Latest stable Python3.x suits best for HDP3.1.5 and RHEL7.9

We are currently using Python2.7 with HDP3.1.5 (PySpark | Spark2.3) and RHEL 7.9.
We are planning to upgrade to Python3.x as Python2.7 support has ended.
If anyone could please help with determining the latest stable Python3.x? version that will best suit or compatible with HDP3.1.5 (Spark2.3) and RHEL 7.9?
OR
Let me know necessary steps I should follow to determine the best stable Python3.x? version for our HDP stack?
Thanks,
PB
The max Python version supported is 2.7.5, Python Version 3.x is not supported as it will create conflict with the versions shipped (HDP3.1.5 + Ambari2.7.5).
Python3 support is not on the road map of HDP 3x and even for Ambari.
This is something which may be done in future to make Python3 as the default for the OS Cloudera supports.

Which version of Miniconda has Python 3.6 for Windows 64-bit?

I'm working on a project which need these deep learning libraries - keras and tensorflow. Unfortunately, these do not work with Python 3.7. Can someone please tell me a Miniconda version (for Windows 64-bit) which has Python 3.6?
I've tried creating virtual environment with Python 3.6, but the libraries don't install properly.
Any help will be appreciated! Thanks!
I wanted to find the last package in the 3.6 series, so I found the announcement of the python 3.7 package build (September 4, 2018) and then looked that package up by date in https://repo.continuum.io/miniconda/ (Miniconda3-4.5.11), then traced it back one release. The end result:
Miniconda3-4.5.4 was the last Python 3.6 miniconda package.
Miniconda3-4.5.11 is the first that uses 3.7.
There are no releases 4.5.5->4.5.10 in that repository.
I just downloaded Miniconda3-4.3.31 for Windows 64bit and it came with Python 3.6.3.
https://repo.continuum.io/miniconda/
Binary: https://repo.continuum.io/miniconda/Miniconda3-4.3.31-Windows-x86_64.exe

Twisted for python3.5?

There is no Twisted for python 3.5 at https://pypi.python.org/pypi/Twisted ,I am using python 3.5.2,what should I do?
I recently encounter a similar problem.
try this Unofficial Windows Binaries for Python Extension Packages
http://www.lfd.uci.edu/~gohlke/pythonlibs/#twisted
Use command "pip install yourfilename.whl"
I used it to fix the problem while installing scrapy
Each new release of Twisted includes updates that bring full equal functionality to the python3 version. Releases are roughly monthly, so you may find that the module already works for you.
It is possible (for example with virtualenv) to install both python2 and 3 How to install both Python 2.x and Python 3.x in Windows 7
Twisted support of python 3 on windows is lagging a little behind support on other systems. So if you have access to another OS you may find that it will allow you to install twisted for python 3.

Will a higher version of Python build Node.js?

I'm running the latest Ubuntu (14.04) and have just downloaded the Node.js tarball. I'm not very familiar with Python though, and although I love Ubuntu, I'm also not familiar with the process of downloading and installing stuff on Linux as well :/
It says on the Node.js download page that Python 2.6 or 2.7 is required, but I have found that newer versions of Python are available. Will a newer version of Python build it correctly?
If you're not familiar with Linux then I don't think you want to compile node.js yourself from source. You can still use the binaries.tar.gz
There are Ubuntu packages that will install node.js for you! Installing from a package manager is easier, try https://github.com/joyent/node/wiki/installing-node.js-via-package-manager#debian-and-ubuntu-based-linux-distributions
As for building from source, there are known issues with Python 3, ref https://github.com/joyent/node/wiki/installation#known-issues, so maybe stick with 2.6 or 2.7
Python versions starting with and including 2.6 up through before 3.0 should work fine. Python 3 brought some backwards incompatible changes with Python 2.x and AFAIK gyp has not been updated to support Python 3 yet.

When Python 3.2.3 isn't good enough

Something I'm doing works under the latest Windows Python 2 and 3, latest Debian Python 2 and 3, and Cygwin Python 2, but fails under Cygwin Python 3.2.3. Known bug fixes after 3.2.3 on Windows deal with the problem.
Where does that realistically leave me concerning deployment, if the latest version using traditional distribution paths (latest Cygwin Python 3) isn't high enough, should I:
tell the user oops, wait for an upgrade
tell the user to manually download a newer Python to shadow the official distro version
tell the user to use Windows Python 3 instead of Cygwin Python 3 to run this program
downgrade my code to stop using the multiprocessing class where bad things happen
I'm testing on six versions of Python right now and didn't even try OS X or BSD yet. Are compatibility issues and having to downgrade code to just not use certain Python Standard Library classes something others have to do often?
I accept the answer from Billy ONeal in the comment above

Resources