基于Lightgbm和XGBoost的优化深度森林算法

谢军飞, 张海清, 李代伟, 于曦, 邓钧予

Optimized deep forest algorithm based on Lightgbm and XGBoost

Junfei Xie, Haiqing Zhang, Daiwei Li, Xi Yu, Junyu Deng

表3 LIGHT?XDF和十种对比算法在八个数据集上的算法召回率比较

Table 3 Model recall rates of LIGHT?XDF and other ten algorithms on eight datasets

数据集 RF NB KNN SVM Lightgbm XGBoost ASTGNN ada⁃mdf lgb⁃df Gcforest LIGHT⁃XDF Chemical Engineering 0.875 0.808 0.249 0.251 0.753 0.875 0.861 0.991 0.875 0.866 0.881 EEE 0.75 0.783 0.353 0.253 0.721 0.901 0.9 0.911 0.729 0.833 0.922 Mechanical Engineering 0.75 0.714 0.327 0.25 0.66 0.833 0.796 0.75 0.842 0.75 0.833 Adult 0.769 0.798 0.777 0.62 0.793 0.798 0.884 0.775 0.792 0.783 0.798 Dry_Bean 0.936 0.762 0.716 0.624 0.934 0.936 0.91 0.913 0.939 0.92 0.94 Bank Marketing 0.683 0.715 0.621 0.73 0.732 0.73 0.697 0.685 0.725 0.695 0.714 Winequality⁃red 0.367 0.371 0.254 0.205 0.36 0.36 0.343 0.284 0.379 0.285 0.373 Winequality⁃white 0.363 0.312 0.223 0.145 0.378 0.393 0.413 0.309 0.362 0.254 0.354