ssb-pkg 0.0.0

A package for detecting and visualizing sinks, sources, and bridges in networks.

General Info:

• Check out https://chipdelmal.github.io/MoNeT/SSBSTP for general project details!

Structure:

• ssb
  • data
    • some_locations.csv
    • kernels
      • kernel_1.csv
      • kernel_2.csv
      • ...
      • kernel_n.csv
  • outputs
  • __init__.py
  • ssbplots.py
  • clusters.py
  • detector.py
  • mpc_defs.py

Package Overview:

• This package provides a more convenient way to perform sink/source/bridge detection for the MoNet project. Given a list of clusters, a list of kernels, and a network of locations, the clusters.py script will iterate over all clusters, and, for each one, run sink/source/bridge detection for each kernel using the specified locations. The script uses multiprocessing to compute the results for large inputs more efficiently. The results are placed in a nested dictionary such that dictionary[c][k] is the detector object used with c clusters for kernel k where k is the file name of the kernel. The package also comes with a plotting library for examining the results of the script in a more appealing way.

Overview of Major Files:

• detector.py : contains the class definition of the detector class. The detector class takes in a state transition matrix (or kernel) and a pandas DataFrame with a 'pop' column and performs sink/source/bridge detection on the data. There are various other settings that may be used with the detector all of which are documented in the file.
• clusters.py : the script for running multi-process clustering detection. The script takes the following flags:
  • -l : the path to the locations CSV file (e.g. 'data/some_locations.csv')
  • -k : a path to a directory of kernels to use for detection. Kernels are expected to be CSV files (e.g. 'data/kernels').
  • -o : the name of the output folder. If unspecified, a folder named 'outputs' will be created (if it doesn't already exist) and the results will be placed in it.
  • -c : a comma separated string of integers representing the number of clusters to use for detection.
  • -n : the number of processes to use. If unspecified then the number returned by cpu_count() is used.
  • -e : the experiment id, which is used for naming the resulting output file.
• mpc_defs.py : contains a helper function for clusters.py. This is needed for multiprocessing.
• ssbplots.py : a module for plotting the results of the clusters.py script. More details about each plot are given within the file.