Issues following RGCN demo #2019
Unanswered
davidshumway
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Following the RGCN demo and therefore attempting to recreate the AIFB dataset. In particular, starting from here: (
stellargraph/stellargraph/datasets/datasets.py
Line 477 in 5ca1e59
Similar to the
affiliation
relationship, I'd like to predict a "hasResult" relationship, which happens to hold a simple result (0-9) for a water quality sample.The issue is that either
nodes_onehot_features = pd.get_dummies(nodes.index).set_index(nodes.index)
or usingnodes_onehot_features
is taking up too much memory. (# nodes: ~2 million, # edges: ~20 million).An alternative to this is adding one feature to each node, simply whether the node is a source or target. However, later on this runs into issue when trying to use the keras
fit
method.Any tips?
As an aside, the demo could be improved by making the affiliation DataFrame simpler to understand and/or refactoring it altogether. From the docs,
affiliation
"is a DataFrame containing the one-hot encoded affiliation of the 178 nodes that have an affiliation."Beta Was this translation helpful? Give feedback.
All reactions