Closed tyler-tomita closed 8 years ago
:ok_hand:
On Wed, Aug 3, 2016 at 4:58 PM, Tyler Tomita notifications@github.com wrote:
- Fig 2
- Absolute error instead of relative
- Report error on unseen test set
- Report time for best single hyperparameter rather than aggregate time for all hyperparameters
- Fig 3
- Pass to ranks for all methods and not just for RerF
- Fig 4
- Replace performance profiles with distributions of classifier rankings
- Pass to ranks for all methods
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ttomita/RandomerForest/issues/85, or mute the thread https://github.com/notifications/unsubscribe-auth/AACjcjKc3vj9zXHNQg-ygvxYYxldXTCCks5qcQD7gaJpZM4JcFr0 .
the glass is all full: half water, half air. neurodata.io, jovo calendar https://calendar.google.com/calendar/embed?src=joshuav%40gmail.com&ctz=America/New_York
you are working on these, right?
Figure 2
Figure 4
fig 3 is still in progress
fig 4 will have a fifth panel for outliers. how do you think the layout of the five panels should be?
we can try 5 across and see how it goes?
On Wed, Aug 31, 2016 at 1:07 AM, Tyler Tomita notifications@github.com wrote:
fig 3 is still in progress
fig 4 will have a fifth panel for outliers. how do you think the layout of the five panels should be?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/ttomita/RandomerForest/issues/85#issuecomment-243659702, or mute the thread https://github.com/notifications/unsubscribe-auth/AACjcu1caNR-KdYa3s_18cf9PS3Nx71oks5qlQwggaJpZM4JcFr0 .
the glass is all full: half water, half air. neurodata.io, jovo calendar https://calendar.google.com/calendar/embed?src=joshuav%40gmail.com&ctz=America/New_York