Ranking

[psychemedia]

One of the things I've been using Gephi for is finding outliers in a cleaning phase of data handling, such as finding nodes with a low value for some variable compared to other nodes.

Whilst I could use filters to separate out these nodes, it struck me that the node size ranking options can be used to do this interactively and visually if we could set min max sizes eg as maxVal minVal limits (ie set the min size limit to be greater than the max size limit); this would result in allowing small values to be visualised as large nodes and large vals as small nodes.

At the moment, I've been manipulating the original dataset outside of gephi, either linearly (dataVal -> -dataVal) or non-linearly (eg dataVal -> 1/dataVal).

On a related point, I wonder whether it might also be useful to be able to plot node size relative to the log of a datavalue?

tony