Fig. 4From: Identifying and understanding determinants of high healthcare costs for breast cancer: a quantile regression machine learning approachEstimated out-of-bag average quantile loss for the 90th percentile of episode expenditures corresponding to each QRFs model, which includes the remaining k variables (numbered by 1, 2, …, k) after sequentially removing variables (numbered by k + 1, …, 11) with lower importance scores, where k = 1, 2, …, 11. The null model is the intercept only modelBack to article page