"Experimental Study of Leaf Confidences for Random Forest" . "Physica Verlag" . . "8"^^ . . "2"^^ . "Decision forests (ensembles of trees) achieve usually smaller generalization error compared to single trees. In the classical methods for growing forests, bagging and boosting, the individual trees are constructed by methods originally developed for growing a single tree as the final predictor. In particular, the trees are usually pruned. For such trees, using weights (confidences) for individual trees improves the accuracy of the prediction of the ensemble. Random forests technique (Breiman 2001) uses a specific tree growing process, which does not produce good individual trees, but the whole ensemble frequently achieves better results than ensembles of trees obtained by classical bagging and boosting. One of the default features of Random Forests technique is that it does not use any weights. The current paper presents experiments demonstrating that in specific situations, appropriately chosen weights may improve the prediction for Random Forests of limited size." . . "Prague" . "Decision forests (ensembles of trees) achieve usually smaller generalization error compared to single trees. In the classical methods for growing forests, bagging and boosting, the individual trees are constructed by methods originally developed for growing a single tree as the final predictor. In particular, the trees are usually pruned. For such trees, using weights (confidences) for individual trees improves the accuracy of the prediction of the ensemble. Random forests technique (Breiman 2001) uses a specific tree growing process, which does not produce good individual trees, but the whole ensemble frequently achieves better results than ensembles of trees obtained by classical bagging and boosting. One of the default features of Random Forests technique is that it does not use any weights. The current paper presents experiments demonstrating that in specific situations, appropriately chosen weights may improve the prediction for Random Forests of limited size."@en . "RIV/67985807:_____/04:00105191!RIV/2005/MSM/A06005/N" . "Kotr\u010D, Emil" . . "2004-08-23+02:00"^^ . "Experiment\u00E1ln\u00ED studie v\u00E1\u017Een\u00ED list\u016F pro metodu Random Forests"@cs . "P(LN00A056)" . . "563749" . "1767;1774" . "RIV/67985807:_____/04:00105191" . "Experimental Study of Leaf Confidences for Random Forest"@en . . "Heidelberg" . . . . "Rozhodovac\u00ED lesy (soubory rozhodovac\u00EDch strom\u016F) obvykle dosahuj\u00ED men\u0161\u00ED generaliza\u010Dn\u00ED chyby ne\u017E jednotliv\u00E9 stromy. V klasick\u00FDch metod\u00E1ch pro konstrukci les\u016F (bagging a boosting) jsou jednotliv\u00E9 stromy konstruov\u00E1ny metodami, kter\u00E9 byly p\u016Fvodn\u011B navr\u017Eeny pro konstrukci jednotliv\u00FDch strom\u016F jako\u017Eto v\u00FDsledn\u00FDch prediktor\u016F. Speci\u00E1ln\u011B, tyto stromy jsou obvykle pro\u0159ez\u00E1v\u00E1ny. Pro takov\u00E9to stromy pou\u017Eit\u00ED vah list\u016F pro jednotliv\u00E9 stromy zvy\u0161uje p\u0159esnost predikce cel\u00E9ho souboru. Metoda Random Forests (Breiman 2001) pou\u017E\u00EDv\u00E1 specifick\u00FD postup konstrukce strom\u016F, kter\u00FD ned\u00E1v\u00E1 dobr\u00E9 jednotliv\u00E9 stromy, ale cel\u00FD soubor \u010Dasto dosahuje lep\u0161\u00EDch v\u00FDsledk\u016F ne\u017E soubory z\u00EDskan\u00E9 klasick\u00FDm baggingem a boostingem. Jedna ze z\u00E1kladn\u00EDch vlastnost\u00ED metody Random Forests je, \u017Ee nepou\u017E\u00EDv\u00E1 \u017E\u00E1dn\u00E9 v\u00E1hy list\u016F. Prezentovan\u00FD \u010Dl\u00E1nek ukazuje v\u00FDsledky experiment\u016F, kter\u00E9 potvrzuj\u00ED, \u017Ee v ur\u010Dit\u00FDch situac\u00EDch vhodn\u011B vybran\u00E9 v\u00E1hy list\u016F mohou zlep\u0161it predikci souboru z\u00EDskan\u00E9ho metodou Random Forests p\u0159i omezen\u00E9m po\u010Dtu pou\u017Eit\u00FDch strom\u016F."@cs . "[D4E86D836E0E]" . "Experimental Study of Leaf Confidences for Random Forest" . . . . . "Computational Statistics" . . . "Savick\u00FD, Petr" . . "decision trees;random forest;weights;leaf confidences"@en . . "3-7908-1554-3" . "Experiment\u00E1ln\u00ED studie v\u00E1\u017Een\u00ED list\u016F pro metodu Random Forests"@cs . . "Experimental Study of Leaf Confidences for Random Forest"@en . "2"^^ .