All source code of "Community Detection using Social Relations and Trajectories".
- Unix-like system with the
/tmp
directory - At least 8 GB of RAM
- clang++, supporting C++17, invokable with
clang++
- The following C++ libraries installed:
- Boost C++ Libraries
- Eigen
- ben-strasser/fast-cpp-csv-parser
- p-ranav/argparse
- pybind/pybind11
- Python 3, invokable with
python3
, with the following extensions:- folium
- haversine
- matplotlib
- more-itertools
- networkx
- notebook (JupyterLab FAILS to display pyecharts visualizations!)
- numpy
- pandas
- pybind11
- pyecharts
- scipy
NOTE: if you need to change experiment parameters, edit parameters.sh
.
unzip social_networks.zip
unzip trajectories.zip
Run bash build_tools.sh
.
Run bash run_experiments.sh
. This process takes a long time (around a day).
Run the following Jupyter Notebooks:
case_study.ipynb
average_trajectory_similarity_calculation_time.ipynb
spatial_temporal_distance_correlations.ipynb
m_values_to_sizes.ipynb
m_values_to_pairwise_distances.ipynb
m_values_to_pairwise_similarities.ipynb
community_detection_times.ipynb