0

For example, I am predicting a score that can have value from 0 to 100. The RMSE = 10. How can I interpret it in layman words? Does it means:

  1. RMSE showed that prediction error calculated with RMSE method was from -5 to +5 from actual value.
  2. RMSE showed that prediction error calculated with RMSE method was from -10 to +10 from actual value.
  3. Some other kind of interpretation?
Stephan Kolassa
  • 123,354
vasili111
  • 1,059
  • RMSE is more useful for model comparisons/model selection that interpreting a single model's ability to predict. Instead you should use mean absolute error (MAE) or mean absolute percentage error (MAPE). This has a much more intuitive interpretation: average distance of of the prediction error (either in absolute terms or percentage terms). – dlnB Apr 29 '20 at 18:36
  • Be careful not to mix up the interpretation of RMSE and the interpretation of MAE (mean absolute loss, so adding up the absolute values instead of the squares). In particular, RMSE does not give you the average amount by which your predictions miss the true values, but MAE does. – Dave Apr 29 '20 at 18:39
  • @dlnB Can you please answer to https://stats.stackexchange.com/questions/463554/how-do-i-interpret-mean-absolute-error-mae-or-mean-absolute-percentage-error question? – vasili111 Apr 29 '20 at 18:41
  • @Dave Can you please answer to https://stats.stackexchange.com/questions/463554/how-do-i-interpret-mean-absolute-error-mae-or-mean-absolute-percentage-error question? – vasili111 Apr 29 '20 at 18:41
  • I added the [tag:rms] tag to your question. Looking through previous questions carrying the tag may be helpful. – Stephan Kolassa Apr 30 '20 at 07:51

0 Answers0