search

dotnet / machinelearning-modelbuilder

Simple UI tool to build custom machine learning models.
Creative Commons Attribution 4.0 International
265 stars 56 forks source link

Multi-output regression using ML.NET #2391

Open asheesh1996 opened 2 years ago

asheesh1996 commented 2 years ago

Hi, flaml has a implementation of doing (Multi-output regression).

Is there any implementation of same available in ModelBuilder , ML.NET CLI or Microsoft.ML.AutoML? If there are any existing examples, please share.

It will be really nice if this can be implemented in Model Builder in Visual Studio.

michaelgsharp commented 1 year ago

@luisquintanilla @JakeRadMSFT @LittleLittleCloud

asheesh1996 commented 1 year ago

Hi, Is this enhancement being planned any time soon?

LittleLittleCloud commented 1 year ago

@luisquintanilla for notification

ASAIK it's not in the next milestone. The major blocking issue is ML.Net doesn't support multi-target regression. If ML.Net support multi-target regression, it won't take too much effort for AutoML.Net or ModelBuilder to leverage it.

For now as a work-around. For a multi-output regression with n targets, you can simply train an AutoML regression model for each target and return all targets as a whole. This is also what sklearn.multiregression does.