Discrepancy in Train and Test Sizes in Various Datasets

[rdyan0053]

I've noticed a pattern in several datasets where the test set size significantly exceeds the train set size. Specifically, in the ECG5000 dataset, the train size is 500 while the test size is 4500. Can someone explain the rationale behind this? Why is the test set so much larger than the training set?

Furthermore, this isn't limited to just the ECG5000 dataset. The FaceAll dataset also exhibits a similar pattern. Are there specific reasons or experimental designs that necessitate such splits?

Looking forward to understanding this better.

Thank you!