Why len(test_edges_false) equals to len(test_edges)？

[CXX1113]

Is there any theoretical support behind this evaluation method?

[bigheiniu]

The balance of positive and negative samples for binary classification. e.g If the sample 90% of edges are positive, the model will learn to predict 1 whatever data it got and achieve 90% accuracy ratio

[CXX1113]

The balance of positive and negative samples for binary classification. e.g If the sample 90% of edges are positive, the model will learn to predict 1 whatever data it got and achieve 90% accuracy ratio

However, this does not reflect the true distribution. If the model is biased against the positive side and used for practical use, it will cause problems, that is, many false positives will be generated.