Metrics¶
netseg currently supports the following metrics:
Metric Name |
Year |
Network Type Original |
Network Type |
Multiple Groups |
Theoretical Complexity |
Range |
|---|---|---|---|---|---|---|
Boundary Connectivity |
2013 |
Undirected |
Both |
Yes |
O(M * (V + E)) |
[-0.5, 0.5] |
Random Walk Controversy |
2018 |
Undirected |
Both |
Yes |
O(M * S * L) |
[-1, 1] |
Dipole Moment |
2015 |
Undirected |
Both |
No (Two Only) |
O(M * I * E) |
[0, 1] |
Freeman’s Segregation Index |
1978 |
Undirected |
Both |
Yes |
O(M * (V + E)) |
[0, 1] |
Krackhardt E-I |
1988 |
Both |
Both |
Yes |
O(M * (V + E)) |
[-1, 1] |
Segregation Matrix Index |
1997 |
Directed |
Both |
Yes |
O(M * (V + E)) |
[-1, 1] |
Coleman’s Homophily Index |
1958 |
Directed |
Both |
Yes |
O(M * K * (V + E)) |
[-1, 1] |
GAM Index |
1989 |
Both |
Both |
Yes |
O(M * (V + E + K²)) |
[\(-\frac{1}{G - 1}\), 1] |
ORWG |
2001 |
Both |
Both |
Yes |
O(M * (V + E)) |
[0, \(\infty\)) |
Spectral Segregation Index |
2007 |
Undirected |
Undirected |
Yes |
O(V + M * V³) |
[0, ∞) |
Assortativity Coefficient |
2003 |
Both |
Both |
Yes |
O(M * (V + E)) |
[\(-\frac{\sum a_i b_i}{1 - \sum a_i b_i}\), 1] |
Note: Complexity notations are defined as follows: V: Vertices, E: Edges, M: Null Models, S: Simulations, L: Walk Length, I: Iterations, K: Unique Groups. V, E, and M are applicable to all metric complexities where listed. Given complexities are adjusted for the worst case for Random Walk Controversy, Dipole Moment and SSI. If null models are not given specifically, M = 1.
Note: netseg implementation is almost always \(V \log V\) instead of \(V\) due to guardrails and sorting of the nodes. Above notation is the metric itself.
Note: netseg extends the measures and allows undirected measures to be applied to directed networks and vice versa. In this case it raises a warning to notify the user.
To propose a new metric for inclusion, please contact the authors with a concise summary of the measure, its computational complexity, your proposed implementation, and a clear explanation of how it uniquely differentiates itself from the existing suite of metrics.