deploy a model from a python notebook
Hello,
Is it possible to deploy a model (sklearn ) for example a regression logistic model trained in a python notebook ?
I am used to go on lab, let the build in model trained and deploy the ultimate model y clicking on "deploy" button. Now, I have a project where I have custom encoding before training and was obliged to work all thin in a python notebook. Is it possible to deploy the python sklearn object? so that when called in the future we do not have to re-train again.
Thanks a lot
Answers
-
There are 2 straightforward ways to go from handmade models in python to visual models usable in other parts of DSS than code:
- Use this code to build a custom python visual model
Use this if you want to let DSS make the preprocessing and want to make your custom estimator from scikit-learn code. - Import as an mlflow model Use this if you want full control over the model
From what you've said it seems you are more in the second case.
Note that if it's "only" about a custom preprocessing, you can write your custom preprocessing code in a regular built-in visual model.
- Use this code to build a custom python visual model
