BoostAODE: Retrain #20

New Issue

rmontanana · 2024-02-26T12:56:53Z

rmontanana commented

2024-02-26 12:56:53 +00:00

Only if convergence: true

When convergence is set to true the train dataset is splitted in two, train & test and only the train part is used to train the model. This suppose that the models trained this way are trained with less information than the models trained with convergence set to false. This proposal tries to ease this problem.

Once the process has ended we have obtained the order in which all the features have been processed and we are going to use this order to make a complete train process using this order to update the models significances. The weights used to train each model in retrain will be 1/m

**Only if _convergence: true_** When _convergence_ is set to _true_ the train dataset is splitted in two, _train_ & _test_ and only the _train_ part is used to train the model. This suppose that the models trained this way are trained with less information than the models trained with _convergence_ set to false. This proposal tries to ease this problem. Once the process has ended we have obtained the order in which all the features have been processed and we are going to use this order to make a complete train process using this order to update the models significances. The weights used to train each model in retrain will be 1/m

rmontanana added this to the v1.0.4 milestone 2024-02-26 12:56:53 +00:00

rmontanana added the

💡 enhancement

label 2024-02-26 12:56:53 +00:00

rmontanana self-assigned this 2024-02-26 12:56:53 +00:00

rmontanana modified the milestone from v1.0.4 to v1.0.5

2024-03-06 16:07:00 +00:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: rmontanana/BayesNet#20