I have generated so many models having a different coefficient of determination (r2) and RMSE. In this analysis, it has been seen that it is not mandatory that a model having maximum r2 has a minimum RMSE. It is very difficult for me to choose a model among them. I just want your suggestion. Which model should I choose? Should I go with r2 or RMSE?
Thank you.
They are two different things... RMSE is a measure of the error without regards to the effectiveness of the model. RMSE can be low and the model could still be worthless.
R2 compares Mean squared error of your model with the results that would be available if only the mean had been used i.e. R2 tells you if your model has better predictions compared to if only the mean had been used as a standard prediction. R2 more than zero means you are doing better than mean, R2 less than zero, your model is worse than just the mean of the labels.
I would go with the R2 and the RSME is a concern for evaluating the model fit. Whereas the R2 relates to the strength of the relationship. So, focusing on the R2 is relevant.
V Raja Sreedharan can you give any article reference? It will be helpful.
- Badges
- Science topic