10 September 2018
While I try to take the cognitive class,
I found an interesting note about the metrics to evaluate a performance of a model, which are
A. Mean absolute error: the easiest way and the most straight forward way to describe an error
B. Mean absolute error: try to gear up the error therefore can add more the sensitivity
C. RMSE: the most preferable metrics because have the same dimension as the samples; hence the error can directly relate to the samples.
D. RSE: not so common use it directly. But more common to use it to Measure the accuracy of R.
In my research, I usually use R, RMSE and Normalized RMSE, a representation of RMSE in percentage while comparing with the full range of the target of the training data.
Hopefully can improve our knowledge.
Regards,
While I try to take the cognitive class,
I found an interesting note about the metrics to evaluate a performance of a model, which are
A. Mean absolute error: the easiest way and the most straight forward way to describe an error
B. Mean absolute error: try to gear up the error therefore can add more the sensitivity
C. RMSE: the most preferable metrics because have the same dimension as the samples; hence the error can directly relate to the samples.
D. RSE: not so common use it directly. But more common to use it to Measure the accuracy of R.
In my research, I usually use R, RMSE and Normalized RMSE, a representation of RMSE in percentage while comparing with the full range of the target of the training data.
Hopefully can improve our knowledge.
Regards,
Komentar
Posting Komentar