Event Title

Comparing Network Sampling Methods

Advisor(s)

Dr. Lulu Kang, Illinois Institute of Technology

Location

Room A121-1

Start Date

26-4-2019 11:25 AM

End Date

26-4-2019 11:40 AM

Abstract

Networks can be used to analyze systems in the real world, however they are often too large for our computers to analyze within a reasonable amount of time. A solution to this is network sampling methods. These are just ways of sampling a smaller “representative” network that we can analyze. Being representative means that the sample retains certain characteristics of the original network. Because there are many characteristics, it means many different things for a sample network to be representative. I looked at three common sampling methods, being random degree node, random edge induced, and snowball sampling, and compared them based on how similar they were to an arbitrary original network for fundamental characteristics, degree and clustering coefficient.

Share

COinS
 
Apr 26th, 11:25 AM Apr 26th, 11:40 AM

Comparing Network Sampling Methods

Room A121-1

Networks can be used to analyze systems in the real world, however they are often too large for our computers to analyze within a reasonable amount of time. A solution to this is network sampling methods. These are just ways of sampling a smaller “representative” network that we can analyze. Being representative means that the sample retains certain characteristics of the original network. Because there are many characteristics, it means many different things for a sample network to be representative. I looked at three common sampling methods, being random degree node, random edge induced, and snowball sampling, and compared them based on how similar they were to an arbitrary original network for fundamental characteristics, degree and clustering coefficient.