Graph aggregated features

If graph information is provided, the dataset will be augmented with additional features.

For each float feature $i$, the sample $F(v)$ will be augmented with the following aggregated features:

• $mean \; { F_i(u) | (v, u) \in Graph}$,
• $max \; { F_i(u) | (v, u) \in Graph}$,
• $min \; { F_i(u) | (v, u) \in Graph}$,

where $F_i(u)$ is a feature $i$ of the sample $u$.