I did follow the training well up to this step in the google jupyter notebook:
train_missing_links = graph.run("""
RETURN id(author) AS node1, id(other) AS node2, 0 AS label
Here it took forever not finished, so I tried it on the sandbox browser but it returned WebSocket fail after long time.
I changed it like this and could get the result within a minute:
MATCH p = (author)-[:CO_AUTHOR_EARLY*2..3]-(other)
Would you help me to run the original code to proceed the training?
From what you are describing, something in the graph went wrong, leading to a transaction that did not complete.
My suggestion is to create a new sandbox/jupyter environment and skip to that exercise.
Please keep in mind that this course uses the older Graph Algorithms Library that is supported with Neo4j 3.5. I strongly recommend that you take this newly-published course which is an update to this course for Neo4j 4.0 and uses the new Graph Data Science Library:
Using a Machine Learning Workflow for Link Prediction: https://neo4j.com/graphacademy/online-training/gds-data-science/
Thank you for giving the tips. I tried several times but it didn't work but only the modified to get its aggregation like count(p) worked.
Now I gave it up and start the new training with Neo4j 4.0.