Getting a negative output for r2_score, means I have overfitted my model with too many hyperparameters than it can handle to train itself.
Generally if your fitted line is worse than just fitting a horizontal line, you will get the r2 score as negative. This simply means your fitted line is very worse.
I’m totally aligned to Manish KC (Coach)’s answer. Complementing, there can be multiple reasons why this issue is happening, among which are:
- Your descriptors (features) are not able to describe the target variable
- You’ve train your model with a batch of your dataset and is not generalizable to all the samples, causing overfitting (only if your model is performing well over the training set), or underfitting (if even over the training set is performing worse than the naivest model).