2006
DOI: 10.1007/s10021-005-0054-1
|View full text |Cite
|
Sign up to set email alerts
|

Newer Classification and Regression Tree Techniques: Bagging and Random Forests for Ecological Prediction

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

11
1,265
2
12

Year Published

2010
2010
2018
2018

Publication Types

Select...
5
3

Relationship

0
8

Authors

Journals

citations
Cited by 1,905 publications
(1,290 citation statements)
references
References 42 publications
11
1,265
2
12
Order By: Relevance
“…2. This characteristic may explain its more intense use in Ubatuba, since this area is more exposed to the ocean than is Caraguatatuba bay, more protected from the forces of the open sea (PIRES-VANIN et al, 1993), due to the presence of São Sebastião Island.…”
Section: Discussionmentioning
confidence: 99%
See 2 more Smart Citations
“…2. This characteristic may explain its more intense use in Ubatuba, since this area is more exposed to the ocean than is Caraguatatuba bay, more protected from the forces of the open sea (PIRES-VANIN et al, 1993), due to the presence of São Sebastião Island.…”
Section: Discussionmentioning
confidence: 99%
“…3) Hermit crab versus shell: An attempt was made to relate the hermit crab's characteristics (sex, dry weight, cephalothoracic shield length and width, propodus length and width) as independent variables influencing the use of the shell through a regression tree analysis. Regression trees are particularly useful in ecological studies for at least two reasons: this method permits the joint use of both qualitative and quantitative variables as predictors (independent variables) and it allows the inclusion of multiple predictors irrespective of whether multicolinearity exists between them (DEATH; FABRICIUS, 2000;PRASAD et al, 2006). regression trees are, therefore, well suited to relate the hermit crab's characteristics to the shell's, because the former often include both qualitative (e.g., sex) and quantitative variables (e.g., size, weight); further, variables such as size, volume and weight are often closely correlated to each other, i.e., are collinear.…”
Section: Samplingmentioning
confidence: 99%
See 1 more Smart Citation
“…A random forest usually consists of a compilation of classification or regression trees (e.g., 1000 trees in a single random forest) to produce more accurate classifications and regressions than single-tree models (i.e., CART) (Liaw and Wiener, 2002). The trees are grown to maximum size without pruning and aggregation is by averaging the trees (Prasad et al, 2006). It selects only the best split among a random subset of variables at each node, but not among the sequence of pruned trees.…”
Section: Performance Of Modeling Methods Usedmentioning
confidence: 99%
“…Then the Random Forest classification defines the final classes according to a rank for most voted trees. This method has been applied in ecological studies (CUTLER et al, 2007;PRASAD et al, 2006) offering powerful alternatives to traditional parametric and semiparametric The algorithm parameters were set after previous tests to improve its effi ciency. We established 100 decision trees to be created without length and pruning performance constraints.…”
Section: Algorithmsmentioning
confidence: 99%