InterSystems Ideas

Have an idea, suggestion, or something that doesn’t work as expected in InterSystems products or services? Share it here on the InterSystems Ideas Portal.

The Ideas Portal is where community members can propose improvements, report bugs, and help influence the product roadmap across InterSystems products and the overall developer experience. 22% of submitted ideas are implemented by InterSystems or members of the Developer Community.

💡 Ideas and bugs are both welcome, no matter how big or small. You can submit feature requests, usability improvements, workflow suggestions, and bug reports. Whether you’re an experienced expert or just getting started, your fresh perspective is valuable.

🛠️ About bugs and fixes. If you have access to InterSystems WRC, please submit bugs there for immediate action. Bug reports submitted through the Ideas Portal are reviewed and tracked, but do not guarantee immediate resolution.

Start by sharing what could be better - the community and our teams will help take it from there.

Add a new idea Subscribe
Status Community Opportunity
Categories InterSystems IRIS
Created by Davi Massaru Teixeira Muta
Created on May 17, 2024

RELATED IDEAS

specify libraries for vectorization or data preprocessing for IntegratedML

During the development of the application, https://openexchange.intersystems.com/package/DNA-similarity-and-classify we missed specifying how Integrated ML would work with the input data, converting it to a specific vectorization of Python sklearn.

We also missed specifying which machine learning algorithm the iris database would use.

Perhaps due to lack of knowledge in the tool, but these limitations made us choose to code a conventional machine learning model programmatically.

  • ADMIN RESPONSE
    Oct 31, 2025

    Thank you for submitting the idea. The status has been changed to "Community Opportunity".

    Stay tuned!

  • Post comment
  • Benjamin De Boe
    Sep 12, 2024

    Depending on which AutoML engine you're working with, there are parameters to control which model is being used. I'm not sure about the level of control offered for feature engineering, but an upcoming "pluggable models" capability will offer a lot more opportunity to customize which ML techniques are used for building your model.

    Reply