extract features from graph

[eigenvivek]

Sprint 1 Goal:

• write functions for group level feature extraction from graphs
• use my algorithms to construct feature matrices (to be used by clustering team)

DoD:

• [x] two good methods for making a feature matrix
• [x] PR into ndmg
• [x] documentation

[eigenvivek]

Previous comments from jovo:

1. networkx/igraph in Python for finding features
2. be specific

[eigenvivek]

Pseudo code! for the edgelist processing pipeline.

input: n edgelists output: n-by-m feature matrix (1) initialize generator that iteratively yields edgelists (2) for each edgelist . (a) extract features (eigenvalues of the Laplacian, eigenvector centrality, etc.) . (b) append feature vector to new feature matrix (3) return feature matrix

[jovo]

@v715 lasso is crappy, i can explain why in class. choose some other feature selection method :)

[jovo]

PR to ndmg for group level stuff