felipebravom
commented
6 years ago Hi Tatiana, Could you please place the same question on our google group? https://groups.google.com/d/forum/emotionintensity
Cheers, Felipe
On Wed, Oct 11, 2017 at 7:45 PM, Tatiana Shavrina notifications@github.com wrote:
Dear organizers!
What is the motivation for choosing such metrics, as Paerson correlation, etc, when there are commonly used, classic metrics for regressor quality estimations - explained variance score, mean squared error, r2 score...? I tried myself to compare the modelling results by classic metrics and by metrics provided - and the ranking of different regression models with two approaches is different.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/felipebravom/EmoInt/issues/1, or mute the thread https://github.com/notifications/unsubscribe-auth/ADmCg3vd2mFYEYd4Ye1loQjRDwwmnGE_ks5srGP5gaJpZM4P0_L7 .
Dear organizers!
What is the motivation for choosing such metrics, as Paerson correlation, etc, when there are commonly used, classic metrics for regressor quality estimations - explained variance score, mean squared error, r2 score...? I tried myself to compare the modelling results by classic metrics and by metrics provided - and the ranking of different regression models with two approaches is different.