東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Methods for analyzing high dimension...

Ding, Beiying.

Linked to FindBook

Google Book

Amazon

博客來

Methods for analyzing high dimensional data: Classification, measurement error model and graph based association measures, with applications to microarray data.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Methods for analyzing high dimensional data: Classification, measurement error model and graph based association measures, with applications to microarray data./
Author:	Ding, Beiying.
Description:	152 p.
Notes:	Source: Dissertation Abstracts International, Volume: 65-05, Section: B, page: 2175.
Contained By:	Dissertation Abstracts International65-05B.
Subject:	Biology, Biostatistics. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3131821
ISBN:	0496790617

Methods for analyzing high dimensional data: Classification, measurement error model and graph based association measures, with applications to microarray data.
Ding, Beiying.

Methods for analyzing high dimensional data: Classification, measurement error model and graph based association measures, with applications to microarray data. - 152 p.

Source: Dissertation Abstracts International, Volume: 65-05, Section: B, page: 2175.

Thesis (Ph.D.)--Harvard University, 2004.

The advances in computational biology have made simultaneous monitoring of thousands of features possible. The high throughput technologies not only bring about a much richer information context in which to study various aspects of gene function and gene products but they also present statisticians with the challenge of analyzing data with large number of covariates and few samples.

ISBN: 0496790617Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Methods for analyzing high dimensional data: Classification, measurement error model and graph based association measures, with applications to microarray data.
LDR:03647nmm 2200313 4500 001 1840354
005 20050721103001.5
008 130614s2004 eng d
020 $a 0496790617
035 $a (UnM)AAI3131821
035 $a AAI3131821
040 $a UnM $c UnM
100 1 $a Ding, Beiying. $3 1928695
245 1 0 $a Methods for analyzing high dimensional data: Classification, measurement error model and graph based association measures, with applications to microarray data.
300 $a 152 p.
500 $a Source: Dissertation Abstracts International, Volume: 65-05, Section: B, page: 2175.
500 $a Adviser: Robert Gentleman.
502 $a Thesis (Ph.D.)--Harvard University, 2004.
520 $a The advances in computational biology have made simultaneous monitoring of thousands of features possible. The high throughput technologies not only bring about a much richer information context in which to study various aspects of gene function and gene products but they also present statisticians with the challenge of analyzing data with large number of covariates and few samples.
520 $a As an integral part of machine learning, classification of samples is often of interest to scientists. In the first chapter, we address classification by extending partial least squares (PLS) in the context of generalized linear regression based on a previous approach, Iteratively ReWeighted Partial Least Squares, i.e. IRWPLS (Marx 1996). We show that by phrasing the problem as a generalized linear model setting and by applying bias correction to the likelihood to avoid (quasi)separation, we get better classification results compared with other classifiers, especially the two-stage PLS regression where categorical outcomes are treated as continuous (Nguyen and Rocke 2002a,b). A feature selection procedure is also examined.
520 $a Distance measures have been used in functional clustering analysis. Pearson correlation may be biased due to inaccuracy in measurement. In the second chapter, we propose a model that not only incorporates measurement error but also allows correlated error distribution. Simulation shows that the model estimates correlation coefficient for true expression with less bias compared with Pearson correlation. There are also optimal characteristics with MSE and mean and variance estimates of the true expression distribution are unbiased. Applications to gene expression dataset show that there is a potential advantage when using this model in comparison with other correlational distances using methods such as multiple probe sets and the graph structure of the Gene Ontology (GO).
520 $a The large amount of meta-data, ranging from sequences and annotations to Gene Ontology and PubMed, is a rich source of information for validating results obtained from various experimental settings. In the third chapter, we mainly address the question of measuring gene-gene association in the PubMed context. We derive measures of association between genes as well as between gene and gene sets based on the gene-gene co-citation graph. We also utilize the natural graph structure induced by PubMed gene-paper affiliation network for discovering interesting publications related with gene(s) of interest as well as finding disease related genes. These tools provide effective ways to further analyze gene expression using PubMed.
590 $a School code: 0084.
650 4 $a Biology, Biostatistics. $3 1018416
650 4 $a Statistics. $3 517247
690 $a 0308
690 $a 0463
710 2 0 $a Harvard University. $3 528741
773 0 $t Dissertation Abstracts International $g 65-05B.
790 1 0 $a Gentleman, Robert, $e advisor
790 $a 0084
791 $a Ph.D.
792 $a 2004
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3131821