Exploring address interpolation using machine learning.
conda env create -f environment.yml
streets.zip contains most of the streets in Berlin. It was generated by Preparation.ipynb and zipped for convenience.
Unzip it:
unzip streets.zip
The Address Reference and Interpolation notebooks work with the data in data/streets folder. Note that github didn't allow an upload of the entire zip file content due to size limit. I should store it on S3.
To generate a different city's street data from Preparation.ipynb, you'll need data from these sources:
- https://www.openstreetmap.org
- https://openaddresses.io
- Street projection data, which can be extracted from pelias interpolation after setting up the address and road network databases described in its README.