This project contains a simple implementation of DBSCAN intended to illustrate how the algorithm works. It was written to go along with my blog post here.
My implementation can be found in dbscan.py.
In scikit-dbscan-example.py, I run both my implementation and the scikit-learn implementation on a dataset and confirm that the resulting labels match.
To improve the performance of my implementation, you would want to use matrix-vector operations to perform the distance calculations (instead of calculating each distance individually in a for-loop).