Search PubMed⌕ Search
Home/Archive/PubMed

PubMed · 10204200

Combining classification trees using MLE.

W D Shannon·D Banks
View original source ↗Export metadata ↓

Abstract

We propose a probability distribution for an equivalence class of classification trees (that is, those that ignore the value of the cutpoints but retain tree structure). This distribution is parameterized by a central tree structure representing the true model, and a precision or concentration coefficient representing the variability around the central tree. We use this distribution to model an observed set of classification trees exhibiting variability in tree structure. We propose the maximum likelihood estimate of the central tree as the best tree to represent the set. This MLE retains the interpretability of a single tree model and has excellent generalizability. We implement an ascent search for the MLE tree structure using a data set of 13 classification trees that predict the presence or absence of cancer based on immune system parameters.

Explore related subjects

ClassificationClinical Trials, Phase I as TopicDecision TreesEffect Modifier, EpidemiologicHumansMeta-Analysis as TopicModels, StatisticalNeoplasmsJournal ArticleResearch Support, Non-U.S. Gov'tResearch Support, U.S. Gov't, Non-P.H.S.

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

W D Shannon, D Banks. 1999-03-30. Combining classification trees using MLE.. https://doi.org/10.1002/(sici)1097-0258(19990330)18%3A6%3C727%3A%3Aaid-sim61%3E3.0.co%3B2-2

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed1999

Toward more meaningful hierarchical classification of protein three-dimensional structures.

Recently, several hierarchical classifications of protein three-dimensional (3D) structures have been published. However, none of them provides any assessment of the validity of a hierarchical representation or test individual clusters contained within. In fact, testing here of published trees reveals that they vary in meaning. Protein structure similarity measures are then assessed in terms of the robustness of the resulting trees for 24 protein families. A meaningful tree is defined as one in which all the clusters are found to be reliable according to a jackknife test. With the use of this criterion, a previously published similarity measure described as a "better RMS" is shown in fact to be usually less suited to protein fold classification than normal RMS after superposition. Here the "best" protein structure similarity measure for hierarchical classification-in terms of that which after clustering produces the highest number of meaningful trees, 20, for the 24 families-is found to be a new one. This measure includes information on the relationship of a distance at a given aligned position in a pair to the rest of the unique distances at that position in a protein family. There are only 2 families of the 24 tested, the globins (3 trees) and Kazal-type serine proteinase inhibitors (21 trees), in which the topology (branching order) of the meaningful 3D structure-based trees is constant. Thus, a new view of protein family sequence-structure relationships is afforded by comparing meaningful trees for each family. More generally, there is a need for care in interpretation of the results of those molecular biology algorithms that force a tree structure on data without assessing its applicability. Proteins 1999;37:20-29.

Classification↗
PubMed1999

Eco Cyc: encyclopedia of Escherichia coli genes and metabolism.

The EcoCyc database describes the genome and gene products of Escherichia coli, its metabolic and signal-transduction pathways, and its tRNAs. The database describes 4391 genes of E.coli, 695 enzymes encoded by a subset of these genes, 904 metabolic reactions that occur in E.coli, and the organization of these reactions into 129 metabolic pathways. The EcoCyc graphical user interface allows scientists to query and explore the EcoCyc database using visualization tools such as genomic-map browsers and automatic layouts of metabolic pathways. EcoCyc has many references to the primary literature, and is a (qualitative) computational model of E. coli metabolism. EcoCyc is available at URL http://ecocyc. PangeaSystems.com/ecocyc/

Classification↗
PubMed1999

Subordinate-level object classification reexamined.

The classification of a table as round rather than square, a car as a Mazda rather than a Ford, a drill bit as 3/8-inch rather than 1/4-inch, and a face as Tom have all been regarded as a single process termed "subordinate classification." Despite the common label, the considerable heterogeneity of the perceptual processing required to achieve such classifications requires, minimally, a more detailed taxonomy. Perceptual information relevant to subordinate-level shape classifications can be presumed to vary on continua of (a) the type of distinctive information that is present, nonaccidental or metric, (b) the size of the relevant contours or surfaces, and (c) the similarity of the to-be-discriminated features, such as whether a straight contour has to be distinguished from a contour of low curvature versus high curvature. We consider three, relatively pure cases. Case 1 subordinates may be distinguished by a representation, a geon structural description (GSD), specifying a nonaccidental characterization of an object's large parts and the relations among these parts, such as a round table versus a square table. Case 2 subordinates are also distinguished by GSDs, except that the distinctive GSDs are present at a small scale in a complex object so the location and mapping of the GSDs are contingent on an initial basic-level classification, such as when we use a logo to distinguish various makes of cars. Expertise for Cases 1 and 2 can be easily achieved through specification, often verbal, of the GSDs. Case 3 subordinates, which have furnished much of the grist for theorizing with "view-based" template models, require fine metric discriminations. Cases 1 and 2 account for the overwhelming majority of shape-based basic- and subordinate-level object classifications that people can and do make in their everyday lives. These classifications are typically made quickly, accurately, and with only modest costs of viewpoint changes. Whereas the activation of an array of multiscale, multiorientation filters, presumed to be at the initial stage of all shape processing, may suffice for determining the similarity of the representations mediating recognition among Case 3 subordinate stimuli (and faces), Cases 1 and 2 require that the output of these filters be mapped to classifiers that make explicit the nonaccidental properties, parts, and relations specified by the GSDs.

Classification↗