Calculates information gain and gain ratio for feature selection in decision tree models.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.