Classification and regression trees

By Wei-Yin Loh. 2011

1,363 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Statistics and Probability, Algorithms & Complexity

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Statistics and Probability

In Algorithms & Complexity

Citations in the corpus

Interpretability via Model Extraction (2017, methodology): “Essentially,T ∗ is constructed greedily as a CART tree [6], except the gain is computed exactly according to P.”
Constructing classification trees using column generation (2018, methodology): “This results in lack of guidance for constructive algorithms (Breiman et al. 1984).”
Machine learning in concrete strength simulations: Multi-nation data analytics (2014, methodology): “This decision tree method constructs classification trees or regression trees depending on the variable type, which may be categorical or numerical [43,44].”
Nonlinear Models Using Dirichlet Process Mixtures (2007, methodology): “We compare our model to a simple MNL model, an MNL model with quadratic terms (i.e., xlxk, where l = 1, ..., p and k = 1, ..., p), referred to as qMNL, and a decision tree model (Breiman et al., 1993) that uses 10-fold cross-validation for pruning.”
A Survey of Random Forest Based Methods for Intrusion Detection Systems (2018, neutral): “In some works (Breiman et al. 1984; Loh 2011), it refers to a graphic representation and the model is referred by “Regression ACM Computing Surveys, Vol. 51, No. 3, Article 48.”
Market Segmentation Trees (2019, methodology): “…set observations for each of the generated datasets, CMTs are trained to depths of 0, 3, and 5, which correspond to 1, 8, and 32 leaves (i.e., market segments), respectively, and we prune the trees using the validation set observations according to the procedure described in Breiman et al. (2017).”
Hubs in Space: Popular Nearest Neighbors in High-Dimensional Data (2010, methodology): “CART trees (Breiman et al., 1984) with the maximal depth of three.24 We define for each pointx its standardized hubnesscore: h(x,k) = Nk(x)−µNk σNk , (24) whereµNk, σNk are the mean and standard deviation ofNk, respectively.”
Recent advances in decision trees: an updated survey (2022, methodology): “One of the main challenges of model trees is efficiency: if the greedy procedure is naively employed, then two predictive models (one for each child node) have to be trained to evaluate the quality of a single candidate split (Loh 2011).”