训练集的 R 平方比测试集的 R 平方大得多。 这对回归 model 意味着什么？

Question

Training Metrics:
R squared: 0.8099510921986353

Testing Metrics:
R squared: 0.17368322884835363

This is the result. I tried optimizing the data but the result does not change by much? What must be wrong with the model or data? What can this difference in value mean for the model and data being used?

Answer 1

Probably your model is overfitted for your training dataset. You could change the hyperparameters or try different resampling methods (eg try lower k for cross-validation).