If you prefer another distribution, you can use your knowledge of that distribution to install Python, NumPy, and others. Here, we will do it on the standard Amazon distribution:

1. We start by installing several basic Python packages as follows:

$ curl -O https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh
$ chmod +x ./Miniconda3-latest-Linux-x86_64.sh
$ bash ./Miniconda3-latest-Linux-x86_64.sh  

2. Now, follow the basic instructions and set the PATH variable as instructed:

$ export PATH=/home/ec2-user/miniconda3/bin:$PATH  

3. Now, we can create a new environment, which we call py3.6 (as it is a Python 3.6 environment) and activate it:

$ conda create -n py3.6 python=3.6 numpy scikit-learn
$ source activate py3.6  

4. To install mahotas and jug, we add the conda-forge channel (it is generally a good idea to do so; it has many well-maintained packages):

$ conda config --add channels conda-forge
$ conda install mahotas imread jug git