1. Mailing Lists
  2. Contributors
  3. Re: New Module "Supervised Learning" - Create Machine Learning Pipelines directly in Odoo

Contributors mailing list archives

contributors@odoo-community.org

Browse archives

Avatar

Re: New Module "Supervised Learning" - Create Machine Learning Pipelines directly in Odoo

by
Invitu, Cyril VINH-TUNG
- 16/02/2025 22:43:14
Hi Michael

I think your contribution will be very welcome in the community and will need new repo(s)...

I suggest connector-ai where your module would be renamed into connector-scikit-learn 
Other modules that will use that connector would go in their respective repo.

Use cases that I can imagine :
- find a similar record in project_task, crm_lead that has been done before and suggest a solution
- suggest automatic entry for supplier invoices in accounting (see Alexis's amazing work here https://m.youtube.com/watch?v=edsEuXVyEYE)
...


Best regards

--------------------------------
Cyril VINH-TUNG
INVITU
Computer & Network Engineering
BP 32 - 98713 Papeete - French Polynesia
Tél: +689 40 46 11 99
contact@invitu.com

Le dim. 16 févr. 2025, 07:37, Elektro-Shop Köck GmbH - Michael Köck <notifications@odoo-community.org> a écrit :

Hello OCA community,

 

I’m excited to share a new module I’ve been working on: Supervised Learning for Odoo. This module provides a flexible framework for creating, training, and deploying classical machine learning pipelines directly within Odoo. It’s still in an early prototype stage (so only a small proof of concept is available at this point), but I’d love to hear your thoughts and ideas for improvement!

 

Why Supervised Learning?

With the current buzz around large language models (LLMs), it’s easy to overlook the benefits of more traditional supervised approaches. In many real-world business workflows, a direct supervised method can be:

 

·       More Accurate for Specific Tasks: Large language models are generalized and can be overkill or less precise for certain domain-specific problems.

·       Resource-Efficient: Training a random forest or logistic regression is typically faster and cheaper than managing an LLM.

·       Interpretability & Control: Classical ML pipelines often let you see feature importances and fine-tune hyperparameters in a more straightforward way, providing better transparency.

·       Privacy & Compliance: With local training on your own Odoo data, you avoid sending sensitive information to external APIs.

 

What the Module Does

·       Integrates https://scikit-learn.org/ into Odoo

·       Allows you to define pipelines consisting of preprocessing, feature extraction, and estimators to make predictions using all the data in your Odoo instance

·       Integrates seamlessly with Odoo models, letting you train and deploy directly in your Odoo instance.

·       Provides a mechanism to add your own custom steps and estimators as needed.

 

Looking for Feedback & Collaboration

Because this is still a prototype, I’d appreciate your feedback on:

1.       Use Cases: What real-world scenarios do you see where such a pipeline would help?

2.       Features: Are there additional estimators or preprocessing steps you’d like to see?

3.       Collaboration: If you have ideas or want to contribute code, let me know! I’d be thrilled to work together on making this more robust.

 

You can find the full source code on GitHub: https://github.com/mkoeck/supervised-learning

I would also be  to integrate this into an existing OCA PCA, but I was unsure which one would fit best. If anyone could guide me in the right direction, that would be great!

 

Thank you for taking the time to check this out. I’m excited to see if others are interested in shaping this into a valuable module for the Odoo ecosystem.

 

Looking forward to your thoughts and suggestions!

 

Best regards,

Michael Köck

_______________________________________________
Mailing-List: https://odoo-community.org/groups/contributors-15
Post to: mailto:contributors@odoo-community.org
Unsubscribe: https://odoo-community.org/groups?unsubscribe