I'm looking to create a machine learning model that could provide, after the prediction, an information about which variable, in this specific case, made the model decide if its prediction was 0 or 1.
I looked some information about that, for example with a logistic regression model, but I only found information about global feature importance (with .coefs for example). What I'm looking at is to be able to explain a specific prediction : For example, explain that a custommer is marked at "1 class" because of "A" variable being high and "B" variable being low, for example. I'm not looking to know if, in general case, "A" and "B" variables have important impact on prediction.
Does such information exist on "common" models, like LogisticRegression ou RandomForest for example ?