Skip to content

Latest commit

 

History

History
28 lines (22 loc) · 2.48 KB

README.md

File metadata and controls

28 lines (22 loc) · 2.48 KB
Raw

TO BE UPDATED

Influence Analysis

This repository contains the code for implementing the algorithms given in Krishnaraj, P. M., Ankith Mohan, and K. G. Srinivasa. "Performance of procedures for identifying influentials in a social network: prediction of time and memory usage as a function of network properties." Social Network Analysis and Mining 7.1 (2017): 34..

To execute the scripts, do the following:

  1. In params.csv, edit the values in column 2, the values after each of the commas.

    1. Do not change column 1, i.e, Input file, File separator, File header, Output path or Seed.
    2. Do not give a space after the comma.
    3. Failure to comply with either of the above steps will result in code failure or abnormal behavior.

  2. Set the current directory to the Influence directory.

  3. Run source("scripts/run.R")

The output will be the following directories:

  1. abscut: An eps file for each of the detected cluster with its influentials identified by the absolute cut score method in red.
  2. cluster: An eps file for each of the detected cluster.
  3. fixcut: An eps file for each of the detected cluster with its influentials identified by the fixed percentage of propulation method in blue.
  4. randhist: A png file for each of the detected cluster with the frequency as function of the in-degree of each of the vertices in this cluster. All vertices with in-degree greater than its random counterpart at 95% level of significance is considered an influential.
  5. randplot: An eps file for each of the detected cluster with its influentials identified by the random permutation method in chocolate.
  6. sample: An eps file for each of the samples until the statistically significant sample is found using the stopping criteria. Each eps file shows the clusters in a different color.
  7. screeplot: An eps file for each of the detected cluster with the screeplot of the indegree of all the vertices corresponding to this cluster:
    1. A vertical blue line for fixed percentage of population method. All vertices to the left of this line are influentials.
    2. A horizontal red line for absolute cut score method. All vertices above this line are influentials.
    3. A horizontal green line for standard deviation method. All vertices above this line are influentials.
  8. sdcut: An eps file for each of the detected cluster with its influentials identified by the standard deviation method in green.