AutoML for Academic Paper

DennisBaloglu

Hello,

I am currently using AutoML for an academic paper and need to write out the some of the underlying mechanisms for the analysis. I was wondering how AutoML works "behind the scenes"? Does it choose the "best" hyperparameters for each analysis? I am using it for classification if that helps.

David_A

Hi,

every step in AutoModel is represented as a RapidMiner operator.

So once you have finished your analysis, you can click onOpen Processand see the underlying RapidMiner process. You can also save the results and use the model to score new data.

As you can see in the processes, there is always a step Train Model -> Optimize in which a hyperparameter tuning takes place.

Best,
David

DennisBaloglu

Got it, so does the optimization change based on the dataset or other factors? Like, how does the optimization itself work? Is it another layer of machine learning?