具有稀疏矩阵数据和多项式Y的xgboost随机森林

I'm not sure if xgboost's many nice features can be combined in the way that I need (?), but what I'm trying to do is to run a Random Forest with sparse data predictors on a multi-class dependent variable.'s many nice features ca