Hello Everyone!
I published some late the challenge, but for this one, I had never worked before with a similar problem, and I needed first to know why was important the data for wine quality.
I appreciate so much the information @MoLa_Data share with me, to understand the requirements for the present challenge, and I based my approach on the solutions published by @MoLa_Data & @alinebessa .
The first was needed clean and adjust the quality features between 4 and 7, this step was made using the Numeric Outliers components.
The next step working on the mean and the standard deviation on the differences between the prediction from the AutoML and the prediction from the node performing Target Shuffling by randomly permuting.
Here are some images from the workflow:
