Normalizing "Clustering Coefficient local"

Submitted by Marieke on

Dear all,

I want to normalize the local CC. In contrast to the global CC and global and local characteristic path length, however, I encounter the following error:

"Normalization error: some metrics derived from random networks result in zero which cannot be used for normalization."

Is there any way I can solve this? When  I look into the MATLAB file, the file truly contains several zeros. How is this possible?

Thank you so much!

Best,

Marieke

 

 

Johann Kruschwitz

Tue, 12/10/2019 - 14:14

Hi Marieke,
the local clustering coefficient  is a measure of local connectedness, measuring the proportion of how many nearest neighbors of node i are connected to each other as well. If all nodes that are connected with node i are not connected among each other a zero can result. It could be the case that your random networks have such properties. Which randomization function do you use? I would suggest using 'c_null_model_und_sign' which is fast and preserves the degree, weight and strength distribution of the original network.
Best,

Johann

Hi Johann,

thank you so much for your quick reply. We used another random network model () but tried c_null_model_und_sign today. We used weighted measures. For the CC, GraphVar notified that we needed binary networks, for the characteristic path length MATLAB crashed.

Is there a way to use weighted networks/measures and normalize these measures?

Thank you so much!

Best

Marieke

Hi Johann,

thank you so much for your quick reply. We used another random network model () but tried c_null_model_und_sign today. We used weighted measures. For the CC, GraphVar notified that we needed binary networks, for the characteristic path length MATLAB crashed.

Is there a way to use weighted networks/measures and normalize these measures?

Thank you so much!

Best

Marieke

Johann Kruschwitz

Fri, 12/13/2019 - 12:41

Hi Marieke,

I think you may have used some incompatible setting in the GUI: if you create weighted random networks and would like to normalize them you would have to select the "weighted clustering coefficient". The warning you described occurs if you create weighted random networks but you chose to calculate the binary clustering coefficienct in "brain graph metrics". Thus, both versions should be identical -> either weighted or binary but not mixed.
Best,

Johann

Forums