Code and data for my paper "Partial Correlation Financial Networks".

I use a virtual env to handle this, there is a requirements.txt that contains the dependencies.
To create the virtualenv
1. virtualenv -p python3 env
2. source env/bin/activate
3. pip install -r requirements.txt 

To obtain the analysis in the paper
1. Run infer_networks.py to create the correlation and partial correlation networks (will also show Figure 3)
2. Copy all the correlation and partial correlation graphml files into seperate folders (i.e. only correlation networks in one folder)
3. In analyze_networks.py file change the networks_folder variable to the folder containing the networks you wish to analyze
4. Run analyze_networks.py to get the figures specific to that network. (Figures 3, 5, 8, 9 and Table 1 can be achieved by this)

For community detection:
1. If you're on Linux, you should be able to do source compile_cython.sh and it'll compile the community detection algorithm - I'm unsure as to how to handle this on windows
2. edit modularity_over_time.py - set the networks_folder variable to the appropriate folder
3. Run it - it will take a long time
4. Change the networks_folder to the other network type
5. Run it again
6. Run community_detection_analysis.py to get Figures 11, 12 and 13.

To obtain Figures 1 and 2:
1. Download gephi - https://gephi.org/
2. Edit the networks_folder variable in display_network to point to the location of the graphml files for the network type you are interesed in
3. Edit the nx.write_graphml call in the threshold_graph method to an appropriate filename (e.g. for the partial correlation networks you could call it partial_correlation.graphml)
4. Run the script
5. Open gephi and import the graphml file in
6. In the appearance tab set the partition to sector and choose your palette
7. Run the Yifan Hu layout
8. You can export this if you wish

To obtain Figures 4 and 6:
1. Open corr_par_corr_comparison.py
2. There are 2 variables to edit here, networks_folder_correlation which contains the location of the correlation networks and networks_folder_partial_correlation
which contains the location of the partial correlation networks
3. Run this script 

Any issues please feel free to contact me.