Learning Model Trees from Data Streams

Ikonomovska, Elena; Gama, João

doi:10.1007/978-3-540-88411-8_8

Cited by 24 publications

(20 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…There are several algorithms to grow linear regression trees and classic regression trees using batch (Breiman et al 1984;Quinlan 1992;Torgo 1997;Dobra and Gehrke 2002) or incremental learning approaches (Potts 2004;Ikonomovska and Gama 2008;Ikonomovska et al 2009). …”

Section: Fuzzy Linear Regression Treesmentioning

confidence: 99%

Fuzzy evolving linear regression trees

2011

View full text Add to dashboard Cite

This paper introduces a new approach for evolving fuzzy modeling using tree structures. The model is a fuzzy linear regression tree whose topology can be continuously updated through a statistical model selection test. A fuzzy linear regression tree is a fuzzy tree with linear model in each leaf. An incremental learning algorithm approach evolves the tree replacing leaves with subtrees that improve the model quality. The learning algorithm evaluates each input only once and do not need to store any past values. The evolving linear regression model is evaluated using time series forecasting problems. The performance is compared against alternative evolving fuzzy models and classic models with fixed structures. The results suggest that fuzzy evolving regression tree is a promising approach for adaptive system modeling.

show abstract

Section: Fuzzy Linear Regression Treesmentioning

confidence: 99%

Fuzzy evolving linear regression trees

2011

View full text Add to dashboard Cite

show abstract

“…To the best of our knowledge, there exists no incremental algorithm for learning multi-target regression or model trees. We then briefly summarize the FIMT algorithm [9], which is the basis of the proposed FIMT-MT algorithm.…”

Section: Related Workmentioning

confidence: 99%

“…The algorithm extends the incremental single-target model tree proposed in [9] by adopting the principles of the predictive-clustering methodology in the split selection criterion. The linear models are computed using a lightweight approach based on incremental training of perceptrons in the leaves of the tree.…”

Section: Introductionmentioning

confidence: 99%

Incremental multi-target model trees for data streams

Ikonomovska

Gama²,

Džeroski

2011

Proceedings of the 2011 ACM Symposium on Applied Computing

Self Cite

View full text Add to dashboard Cite

As in batch learning, one may identify a class of streaming real-world problems which require the modeling of several targets simultaneously. Due to the dependencies among the targets, simultaneous modeling can be more successful and informative than creating independent models for each target. As a result one may obtain a smaller model able to simultaneously explain the relations between the input attributes and the targets. This problem has not been addressed previously in the streaming setting. We propose an algorithm for inducing multi-target model trees with low computational complexity, based on the principles of predictive clustering trees and probability bounds for supporting splitting decisions. Linear models are computed for each target separately, by incremental training of perceptrons in the leaves of the tree. Experiments are performed on synthetic and real-world datasets. The multi-target regression tree algorithm produces equally accurate and smaller models for simultaneous prediction of all the target attributes, as compared to a set of independent regression trees built separately for each target attribute. When the regression surface is smooth, the linear models computed in the leaves significantly improve the accuracy for all of the targets.

show abstract

“…Regarding memory management, Maloof and Michalski [5] suggested that learning algorithms follow one of these three possibilities for the memory model when dealing with past training examples: No instance memory, in which the incremental learner retains no examples in memory but instead statistics about them (e.g., FIMT and FIRT-DD algorithms [6], [7], Online-RD/RA algorithms [8], and neural networks build this kind regression models), Full instance memory, in which the method retains all past training examples (e.g., M5 [9], CART [10], and IBk [11] are regression algorithms following this approach), and Partial instance memory, which is a strategy mainly oriented to dealing with changes in underlying patterns, and learning algorithms retain some of the past training examples that are within a window. The size of this window may be fixed or adaptive.…”

Section: Related Workmentioning

confidence: 99%

Self-Adaptive Induction of Regression Trees

Fidalgo-Merino

Núñez

2011

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

A new algorithm for incremental construction of binary regression trees is presented. This algorithm, called SAIRT, adapts the induced model when facing data streams involving unknown dynamics, like gradual and abrupt function drift, changes in certain regions of the function, noise, and virtual drift. It also handles both symbolic and numeric attributes. The proposed algorithm can automatically adapt its internal parameters and model structure to obtain new patterns, depending on the current dynamics of the data stream. SAIRT can monitor the usefulness of nodes and can forget examples from selected regions, storing the remaining ones in local windows associated to the leaves of the tree. On these conditions, current regression methods need a careful configuration depending on the dynamics of the problem. Experimentation suggests that the proposed algorithm obtains better results than current algorithms when dealing with data streams that involve changes with different speeds, noise levels, sampling distribution of examples, and partial or complete changes of the underlying function.

show abstract

Learning Model Trees from Data Streams

Cited by 24 publications

References 8 publications

Fuzzy evolving linear regression trees

Fuzzy evolving linear regression trees

Incremental multi-target model trees for data streams

Self-Adaptive Induction of Regression Trees

Contact Info

Product

Resources

About