Model Evaluation

What is R-squared value and RMSE and which of them should be near to 0?

RMSE is the error from absolutely correct predictions so it must be close to 0 or as low as possible.

Any type of error must be minimized as much as possible.

It is the R-square value which should be close to 1 for a great model.

But you can't expect your model to achieve such performance-based on so few data points and there might be some problems with the dataset or data collection as well. Try to improve the R2 value as much as you can so that it becomes closer to 1. Or try to decrease the RMSE score as much as you can.