Node Set Influence Analysis Question

[dsprice]

I am trying to determine the set of nodes within my network that influence a given percentage of the entire network within a given distance. Using Eigenvector Centrality and Modularity, I can back myself into what seems like a solution. But my guess is that there's a specific type of analysis available that would be much better.

For example, I've identified a list of performance measurements for an automobile as "outputs" and connected those to a network of influencing "inputs". Modularity allows me to see Neighborhoods of performance connections and Eigenvector Centrality ranks importance, but what I would like to determine is the set of performance inputs with the greatest influence on 90% of the entire network within a distance of 2 from any node in the set of performance inputs.

Any suggestions for a more rigorous analysis path?

Thanks,
David