Hi @alinebessa,
I think for some of us, the choice of a specific model was guided by EDA (done here or on the previous part of the classification challenge), experience/knowledge about algorithms, and more important the “note 1” requirement to only do a workflow with 5 nodes.
Technically speaking, using the AutoML component make the number of nodes rising (much more than 5 ), even if only 5 are present on the main screen.
That may explain why some of us have chosen a specific model, and not the AutoML proposed component.
Looking forward to the next challenges !