In the beginning there was the big-bang, that created everything and all machine learning ideas and algorithms. Now comes the small-bang, a reimplementation of (nearly) everything.
Just git clone and run python3 -m pip install -r requirements.txt
Just run whatever model you want to test, by default invocating it directly will run a simple test.
Make sure you run it from its root directory, for example, to run the K-NN algorithm, use cd knn; python3 knn.py
Since I don't wan't to completely reinvent the wheel, I am frequently using functions from numpy, scipy and scikit learn. These are mostly things to make sure all the heavy numerical processing is done by vectorizing numpy code, alongside loading datasets that come built-in in scikit-learn.