This project gives an implementation of a cluster tracking algorithm used a perform stable clustering throughout a temporal domain.
An implementation of three variations is given in cluster_tracker.py.
We evaluated the different implementations and measured both accuracy and execution times of each approach.
The code used for the evaluation is stored in delta.py.
The data used for the evaluation (multi_days_{lat,lon}.nc3 and single_day_{lat,lon}.nc3 by Rene Schubert) are licensed under CC-BY-4.0.
The file plot_data.py contains the code to plot all data for four consecutive days.
This includes the raw data, as well as the clusters and tracked clusters.