Set up a new virtual environment using conda. We strongly recommending naming the environment the same thing as your project root folder so that it is easy to remember.
Once the environment is created, activate the environment.
When you're finished your command line prompt should look something like:
(water-pumps)machine:~ user$
Use the conda env list command to list the environments that are available and confirm that the water-pumps environment is marked with a * to indicate it is active.
In [ ]:
#SOLUTION
# AT THE COMMANDLINE:
> conda create -n water-pumps python=3
> source activate water-pumps
> conda env list
Now that you have a virtual environment, let's populate it with the packages we need. For convenience, we've listed the some of the contents of your requirements.txt file. If you see an import package error later in the tutorial, you'll be expected to update requirements.txt with the additional dependency!
At this step you need to:
requirements.txt file. It should have been created by the data-science-cookiecutter process.requirements.txt with the packages you need for analysis. Remember, the point is not to pre-load the environment with every possible package you could ever need. Just your minimum reproducible environment.Here's a start for what your requirements.txt should contain.
In [ ]:
#SOLUTION
# AT THE COMMANDLINE:
> cd water-pumps
> edit requirements.txt
> pip install -r requirements.txt
Awesome! Finally, let's add the watermark extension so we can keep track of the package versions every time we execute the notebook. This helps people who find the notebook in other contexts (nbviewer, as a blog post, as a rendered notebook) also know what versions of packages the code executes against. The watermark documenation may help here.
In [ ]:
#SOLUTION
%load_ext watermark
%watermark -a "Peter Bull" -d -t -v -p numpy,pandas,matplotlib -g