Predict apartment prices in Stockholm using machine learning!

Apartment prices in Stockholm. All apartment features are the same on the map, except the position.

Price change for some apartments as function of time.

Price per m^2 for some apartments as function of apartment size.

Price per m^2 for some apartments as function of floor number.

Price per m^2 for some apartments as function of construction year.

• View the figures in the figures folder.

• Install libraries:

./install.sh

• Activate Python virtual environment and set PYTHONPATH.

source env.sh

• Load a (trained) neural network model, and generate a map of Stockholm with apartment prices color-coded, as well as two videos of how the prices in the region have changed over time:

./use_model_tf.sh

• Edit the python file use_model_tf.py to something you are interested in, e.g. add an apartment to be analyzed.

Check unit-tests with

pytest

Check style and types with

make check

The documentation of this package is found in the directory doc.

To update the manual, type:

make -s -C doc/sphinx clean
make -s -C doc/sphinx html

to generate a html-page.