The art of train and test split selection can make or break our understanding of data.

Try this: Experiment with different splits to find the one that aligns with your research question.