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

Statistical models for analysis of h...

Guo, Shyh-Forng.

Linked to FindBook

Google Book

Amazon

博客來

Statistical models for analysis of herd life and prolificacy traits in swine.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Statistical models for analysis of herd life and prolificacy traits in swine./
Author:	Guo, Shyh-Forng.
Description:	180 p.
Notes:	Source: Dissertation Abstracts International, Volume: 67-06, Section: B, page: 2888.
Contained By:	Dissertation Abstracts International67-06B.
Subject:	Agriculture, Animal Culture and Nutrition. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3222915
ISBN:	9780542754203

Statistical models for analysis of herd life and prolificacy traits in swine.
Guo, Shyh-Forng.

Statistical models for analysis of herd life and prolificacy traits in swine. - 180 p.

Source: Dissertation Abstracts International, Volume: 67-06, Section: B, page: 2888.

Thesis (Ph.D.)--The University of Wisconsin - Madison, 2006.

Multiple-trait models are required for estimation of genetic correlations between traits and for increasing accuracy of genetic evaluation. In nucleus herds, censoring of records arises because sows are sold to other herds for production purposes. A multiple-trait linear model allowing for censored records was developed to estimate genetic parameters in a joint analysis of herd life and litter size. Data consisted of records on 2,616 purebred Landrace daughters of 301 sires. Heritability estimates were low, and unfavorable negative correlations were found between herd life and litter size traits. Genetic correlations between different measures of herd life were large. A Poisson mixed effects model was developed for joint analysis of lifetime prolificacy and litter size. Censored records were treated via the data augmentation method. Favorable positive genetic correlations were found between lifetime prolificacy and litter size. A Poisson-t model was compared with a Poisson-normal model for the extra-Poisson variation. Rank correlations between sire evaluations from the two models were large. The Bayes factor and the degrees of freedom estimate did not give strong support to the t-distribution specification. A Poisson random regression model was also designed for genetic analysis of litter sizes in successive parities. A t-distribution was used for extra-Poisson variation, and polynomial functions of farrowing age up to second order were used for the regressions. Genetic variation and heritability were found to increase with age. The small second order regression and the Bayes factor suggested that a linear regression would be adequate. A multiple-trait hazards model with random frailties was developed for joint analysis of two survival traits. Multivariate normal distributions for the logarithm of the random frailty variables introduced correlation structures. Most of the fully conditional distributions did not have closed forms. A Gibbs-Metropolis-Hastings algorithm was described for drawing inferences from marginal distributions. In general, the models were implemented successfully in a Bayesian context. However, estimates of genetic parameters were imprecise and no clear genetic relationship between herd life and litter size emerged from the analysis conducted in this population.

ISBN: 9780542754203Subjects--Topical Terms:

1017857
Agriculture, Animal Culture and Nutrition.

Statistical models for analysis of herd life and prolificacy traits in swine.
LDR:03225nmm 2200277 4500 001 1828942
005 20071023113057.5
008 130610s2006 eng d
020 $a 9780542754203
035 $a (UMI)AAI3222915
035 $a AAI3222915
040 $a UMI $c UMI
100 1 $a Guo, Shyh-Forng. $3 1917817
245 1 0 $a Statistical models for analysis of herd life and prolificacy traits in swine.
300 $a 180 p.
500 $a Source: Dissertation Abstracts International, Volume: 67-06, Section: B, page: 2888.
500 $a Adviser: Daniel Gianola.
502 $a Thesis (Ph.D.)--The University of Wisconsin - Madison, 2006.
520 $a Multiple-trait models are required for estimation of genetic correlations between traits and for increasing accuracy of genetic evaluation. In nucleus herds, censoring of records arises because sows are sold to other herds for production purposes. A multiple-trait linear model allowing for censored records was developed to estimate genetic parameters in a joint analysis of herd life and litter size. Data consisted of records on 2,616 purebred Landrace daughters of 301 sires. Heritability estimates were low, and unfavorable negative correlations were found between herd life and litter size traits. Genetic correlations between different measures of herd life were large. A Poisson mixed effects model was developed for joint analysis of lifetime prolificacy and litter size. Censored records were treated via the data augmentation method. Favorable positive genetic correlations were found between lifetime prolificacy and litter size. A Poisson-t model was compared with a Poisson-normal model for the extra-Poisson variation. Rank correlations between sire evaluations from the two models were large. The Bayes factor and the degrees of freedom estimate did not give strong support to the t-distribution specification. A Poisson random regression model was also designed for genetic analysis of litter sizes in successive parities. A t-distribution was used for extra-Poisson variation, and polynomial functions of farrowing age up to second order were used for the regressions. Genetic variation and heritability were found to increase with age. The small second order regression and the Bayes factor suggested that a linear regression would be adequate. A multiple-trait hazards model with random frailties was developed for joint analysis of two survival traits. Multivariate normal distributions for the logarithm of the random frailty variables introduced correlation structures. Most of the fully conditional distributions did not have closed forms. A Gibbs-Metropolis-Hastings algorithm was described for drawing inferences from marginal distributions. In general, the models were implemented successfully in a Bayesian context. However, estimates of genetic parameters were imprecise and no clear genetic relationship between herd life and litter size emerged from the analysis conducted in this population.
590 $a School code: 0262.
650 4 $a Agriculture, Animal Culture and Nutrition. $3 1017857
650 4 $a Biology, Veterinary Science. $3 1021733
690 $a 0475
690 $a 0778
710 2 0 $a The University of Wisconsin - Madison. $3 626640
773 0 $t Dissertation Abstracts International $g 67-06B.
790 1 0 $a Gianola, Daniel, $e advisor
790 $a 0262
791 $a Ph.D.
792 $a 2006
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3222915