Envoyer la notice par courriel: Weighted Quantile Regression Forests for Bimodal Distribution Modeling: A Loss Given Default Case