Author | Message |
---|---|
306264301
Posts: 28
|
Posted 10:25 Mar 14, 2020 |
I have a question. On the last part of the homework. Do we use gridSearch again, or just cross validation? |
mpourhoma
Posts: 39
|
Posted 11:41 Mar 14, 2020 |
If you use sklearn Gridsearch, it will automatically use cross validation. Please see Lab3,4 tutorials.
|
306264301
Posts: 28
|
Posted 12:58 Mar 14, 2020 |
I see, but this time we apply it on the testing data instead? With the best batch_size and epochs, we got from part g. |
mpourhoma
Posts: 39
|
Posted 13:42 Mar 14, 2020 |
In part g, You have to try it on training data only: grid.fit(X_train, y_train) Last edited by mpourhoma at
18:15 Mar 14, 2020.
|
306264301
Posts: 28
|
Posted 14:15 Mar 14, 2020 |
For part h? |
mpourhoma
Posts: 39
|
Posted 18:18 Mar 14, 2020 |
In part g, You have to try Gridsearch on "training data" to find the best model (associated with the best epochs and best batch-size). Then, in part h, you will evaluate this best model on the testing data. |
306264301
Posts: 28
|
Posted 18:20 Mar 14, 2020 |
Thanks. |