|
From: | Qunawei Zhang |
Subject: | Re: [igraph] question about average.path.length(graph, directed=TRUE, unconnected=TRUE) |
Date: | Sun, 06 Sep 2015 09:58:33 -0400 |
User-agent: | Microsoft-MacOutlook/14.4.6.141106 |
Hello: Thanks. And you think the possible "real" path length within the graph is the number of nodes in the network (minus 1). I just thought that may be it is too stringent. One unconnected node may contribute much to the average length of shortest path (consider one case, only a few nodes are not connected to the connected component). If the diameter of the connected component is d, how about to define the distance between unconnected nodes as d+1 or 2d? Thanks again Best Quanwei From: Tamas Nepusz <address@hidden> Reply-To: Help for igraph users <address@hidden> Date: Sunday, September 6, 2015 9:40 AM To: Help for igraph users <address@hidden> Subject: Re: [igraph] question about average.path.length(graph, directed=TRUE, unconnected=TRUE) Hello, What else would you do with them? :) Obviously we cannot count them with an infinite distance because that would make the average distance infinite as well. We cannot ignore them either, because a disconnected graph with 1 million vertices and a single edge would have an average path length of 1 if we did that. So, the best we can do is to take them into account with a distance that is larger than any possible "real" path length within the graph. T. T. On Sun, Sep 6, 2015 at 2:37 PM, Qunawei Zhang <address@hidden> wrote:
|
[Prev in Thread] | Current Thread | [Next in Thread] |