I am trying to install numpy on openshift so that my django website can access it. I tried to install it by calling pip install numpy in the pod terminal. This seemed to install but when I tried to import numpy in my code it gave ImportError: No module named numpy
If you are using the OpenShift S2I builder for Python, you would be pointing it at your repository with your application source code. That repo should have a requirements.txt file in it which lists the names of the Python packages you want installed. So create that file and add numpy to it and then trigger a new build and deployment of your application.
If you are not using the Python S2I builder, you need to explain how you are running your application under OpenShift.
I have tried installing scipy in Google Cloud Shell. The package is installed, but a python import is giving "ImportError: No module named scipy". Screenshot
I have problem only with scipy. Tensorflow and numpy are all working fine.
This problem should be similar to Installed packages disappeared in Google Cloud Shell.
The problem here is that scipy has not been installed. The process got killed at 99%. This is caused by how pip tries to install the package. I suspect that you're using a small VM instance which has memory limitations when pip tries to load the whole file to memory before installing it.
The solution is to install scipy with this command:
pip --no-cache-dir install scipy
Here you ask pip not to cache the file which should do the trick to install scipy on your Google Cloud VM. After the successful installation you should be able to import the scipy module as intended.
I have a project created in Pycharm with anaconda3 (python3) as root, and it doesnt recognize certain packages (for example in this case Seaborn package). When I try to install from terminal it says, Seaborn is already installed in anaconda2 (python2.7).
How can I manage two different versions of anaconda, because I still have to work with some legacy (python2.7) codes.
My project interpreter is already set to 3.5, whereas the terminal doesn't change accordingly.
Your situation is an ideal case for the usage of virtual environments in python.
Virtualenvs allow you to maintain versions of python without all of the dependency linking etc that it looks like you are getting into.
With Anaconda, this process is even easier as you can use the anaconda's built in manager to create a separate python 2 environment with something like the below.
conda create --name <yourenvname> python=2.7 anaconda
which will install a full blown anaconda environment in 2.7. Doing so will allow you to switch between your python 2 and python 3 without having these managment issues.
For more information see here in Anaconda on using 2or3
I am doing some Python 2.7 development work on multiple computers, on some of which I do not have rights to install software. Thus, I am using Portable Python. One of the packages I need to use is h5py; however, since Portable Python installs without modifying the registry, the h5py installer doesn't see it as a valid Python installation. Also, simply extracting the h5py source to a custom F:\py\include folder and adding F:\py\include to the PYTHONPATH environment variable doesn't work, as import h5py results in an ImportError: Cannot import name _errors exception.
How do I install h5py with Portable Python 2.7?
Install Python 2.7 onto a computer for which you have sufficient administrative rights.
Install the appropriate version of h5py into this Python installation.
Find the h5py subfolder within the Python install (typically .\Lib\site-packages\h5py) and copy it to the F:\py\include folder mentioned (make sure to add F:\py\include to PYTHONPATH). See my answer here for more details.
Uninstall the 'official' version of Python, if desired.
I followed these instructions to set up GraphLab on my Ubuntu machine. At the end, I opened Python 2.7.6 and ran the first of the test lines import graphlab as gl. This gave me
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: No module named graphlab
How can I begin to diagnose this?
Details:
I ran python -V from a terminal, and it returned me Python 2.7.6.
In /usr/bin I find the following pyth* entries ... I wonder if something somewhere pointed at the wrong version:
python python2.7-config python3.4 python-config
python2 python2-config python3.4m pythontex
python2.7 python3 python3m pythontex3
The Dato Graphlab Create installer did not actually install graphlab on my Mac (El Capitan). I did the following (Anaconda is installed) in a terminal window:
% pip install graphlab-create
That subsequently installed Graphlab Create. You can then easily verify:
% python
Python 2.7.10 |Continuum Analytics, Inc.| (default, Sep 15 2015, 14:29:08)
[GCC 4.2.1 (Apple Inc. build 5577)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://anaconda.org
>>> import graphlab
>>>
I've noticed that occasionally, Python will forget that Graphlab Create is installed. A repeat of the above 'pip' command will cause it to remember.
python anaconda graphlab
Another option is to use virtualenvwrapper for the easy creation and application of virtual environments. For example, following this documentation, start with installation:
sudo pip install virtualenvwrapper
Open your .bashrc settings file, for example run gedit .bashrc and append the following lines to the bottom of it:
export WORKON_HOME=$HOME/.virtualenvs
export PROJECT_HOME=$HOME/Devel
source /usr/local/bin/virtualenvwrapper.sh
Restart your terminal window, and then you can make your virtual environment, say call it "test":
mkvirtualenv test
Now test is a virtual environment, and your are in it (i.e., test is "activated" currently). To put GraphLab in test,
pip install graphlab-create
Similarly, you can install other python toolkits in test by using pip, and any python program you run from within test will be able to see only the python toolkits that are installed here.
Maybe you should install graphlab in virtualenv.
1.Ensure your system has virtualenv installed. To verify, execute pip freeze. To install, execute sudo pip install virtualenv in your terminal before proceeding
2.Copy and execute the following commands in your terminal. This will create a virtual environment called 'graphlab' and install graphlab create version 0.9.1
virtualenv graphlab
. graphlab/bin/activate
pip install graphlab-create==0.9.1
You may need to activate the conda env by running
source activate dato-env
inside the terminal
Check your system path
import sys
print sys.path
It should contain graphlab-0.9.1. If not, then something was odd with our installation. I recommend using a virtual environment in python.
I had the same problem on ubuntu 16 desktop. The solution for me was pretty simple. After you start the notebook using
(gl-env) davis#smeagol:~/progs/ml-foundations$ jupyter notebook
Click the file navigator to locate your notebook where you do the import graphlab which causes the error. When it starts the notebook I imagine you see |Python [Root] in the top right. To fix this, click the title bar Kernel->Change kernel->gl-env. Now the top right label should say |Python [gl-env]. Afterwards when you run the notebook import graphlab will work.
There is a tab on the intial landing page of the Jupyter UI which has Conda. In that you can see two env's named root and gl-env. I've tried to delete the root one and even though its not the default all my notebooks start up with that environment and deleting it causes an internal error.
Graphlab is not supported on python3. Install Python 2.7 as mentioned in
https://conda.io/docs/user-guide/tasks/manage-python.html
If you don't see graphlab, simply the path of the environment is not set on "dato-env" (rather it may be set on "root")
If you use "Launcher" application, on top left set "Environment" to "dato-env".
Well,I guess the thread is dead.
After tinkering w/ un/reinstallations a couple times, the only way I can get "import graphlab" to work reliably is to manually activate dato-env.
Open your terminal and type below command
source activate dato-env
Prior to this close all the jupyter notebook. I ascertain that dato-env is in effect when my bash prompt changes to: (dato-env) pydev#smruti:~$
Now on your Jupyter notebook try to do import graphlab,this will execute without showing import error.
Hope this helps!!
I had the same problems, but then I found that in the files that come along with the Machine Learning specialization (https://www.coursera.org/learn/ml-foundations/notebook/lGQH5/open-your-notebook-workspace-to-follow-along) there are some additional codes after which you don't get any errors:
import graphlab
Set product key on this computer. After running this cell, you will not need to re-enter your product key.
graphlab.product_key.set_product_key('your product key here')
Limit number of worker processes. This preserves system memory, which prevents hosted notebooks from crashing.
graphlab.set_runtime_config('GRAPHLAB_DEFAULT_NUM_PYLAMBDA_WORKERS', 4)
Output active product key.
graphlab.product_key.get_product_key()
I had the same problem. I follow these steps.
1.Install Anaconda 2.7 version. Then I created vitual environment and selected python 2.7 version.
2.After create virtual environment open terminal in and run pip install notebook.
3.Then I registered https://turi.com/ because Graphlab Create requires an academic license to use.Run the following command that is given by after registration in terminal.
pip install --upgrade --no-cache-dir https://get.graphlab.com/GraphLab-Create/2.1/your registered email address here/your product key here/GraphLab-Create-License.tar.gz
4.Run jupyter notebook.
5.import graphlab
6.Then I got an error.So i run graphlab.get_dependancies() command. After that restarted the kernel.
7.After above all steps I typed import graphlab again.
8.It executed without errors.