About a bank prediction model (A Machine Learning prediction), we have a feature called "credit_category, it can be good

[answerhappygod]

About a bank prediction model (A Machine Learning prediction),
we have a feature called "credit_category, it can be good, bad or
average. Since our bank is very risk averse, it tries to minimize
the number of credits that we call "good" but actually are "bad".
What is the best performance metric as stated below should we
select? (FOR DECISION TREE) Accuracy or Recall or Precision.
Please provide a detailed explanation, Thank You.