All data to train after evaluating a model

After rigorously training and evaluating a model, is it a good idea to retrain the model with all the data in training to deploy a model in an application with few data?

Hi @hcf2
please can you you better explain what do you mean with
“to deploy a model in an application with few data”?
Anyway I haven’t ever heard about this retrain with all the data into the train dataset.

Indeed, I’m trying to say that the application I want to deploy is poor in amount of data.

Hi @hcf2
If you have just a poor bunch of data your model could be highly inaccurate (high bias) and it could not be deployable in production.
In order to properly train your model you might try to use data augmentation so that the dataset can have a good coverage of the features domain.
If the data augmentation is not a doable approach I would try to clean the data to get the best you can from the available data.
Best regards