LRQMM (version 1.2.2)

lrqmm_m: Fitting Linear Quantile Regression Mixed Models With Relationship Matrix With MATLAB

Description

Fit a quantile regression mixed model involved Relationship Matrix using a sparse implementation of the Frisch-Newton interior-point algorithm.

Usage

lrqmm_m(id, sire, dam, X, Y,cova=NULL , alpha = 0 , tau = 0.5 , Factor = FALSE
, maxTries = 3000, interval = 30)

Arguments

id

The number form animal record as column matrix

sire

The number form father's animal record as column matrix

dam

The number form mother's animal record as column matrix

X

fixed effect(s) as column matrix

Y

a response column matrix

cova

covariate effect(s) column matrix

alpha

a parameter for raite error's varince to variance of random effects, dependent on statistical model (Animal model, Sire model, etc.)

tau

desired quantile

Factor

type of fixed effect that "TRUE" as factor variable and "FALSE" as quantitative variable

maxTries

The maximum number of times the connection is check for an answer from the MATLAB server before giving up. Default values is 3000 times.

interval

The interval in seconds between each poll for an answer. Default interval is 30 (second).

Value

Fixed effects

estimate for fixed effect(s) from linear quantile regression mixed model with its standard error

cova effects

estimate for covariate effect(s) from linear quantile regression mixed model with its standard error

Random effects

estimate for random effect(s) from linear quantile regression mixed model with its standard error

residuals

estimate for model residuals from linear quantile regression mixed model

Time_between_start_to_end

execution time of linear quantile regression mixed model

summary

reporting quantile for effects estimation, mean absolute error for fitted model, variance of response variable, variance of pedigree's random.effect, variance of record's random.effect, number of observations, pedigree's length, fix effect lavels and random effect lavels

Details

The function computes an estimate on the tau-th quantile effects of the linear mixed model. This is a sparse implementation of the Frisch-Newton algorithm for quantile regression described in Portnoy and Koenker (1997).

We used "GeneticsPed", "Matrix", "kinship2", "MCMCglmm", "R.matlab", "SparseM" and "quantreg" packages in this function. befor using "lrqmm" function be sure from installation this packages.

"GeneticsPed" available in

<https://bioconductor.org/packages/release/bioc/src/contrib/GeneticsPed_1.46.0.tar.gz> or orders in <http://bioconductor.org/packages/release/bioc/html/GeneticsPed.html>.

other packages are available in CRAN.

References

[1]Alavian, S. R. (2019). Creating LRQMM package for predicting the breeding value of animals by corrected mixed quantile regression (Unpublished master's thesis). Ferdowsi University Of Mashhad. Mashhad. Iran.[Persian].

[2]Koenker, R. and S. Portnoy (1997). The Gaussian Hare and the Laplacean Tortoise: Computability of Squared-error vs Absolute Error Estimators, (with discussion). Statistical Science, 12, 279-300. <https://www.jstor.org/stable/2246216>

[3]Koenker, R. W. (2005). Quantile Regression, Cambridge U. Press. ISBN: 0521608279.

[4]Mrode, R. A. (2005). Linear Models for the Prediction of Animal Breeding Values. 3rd edition. CABI International. ISBN: 9781780643915.

Examples

# NOT RUN {
#Start(not run)
#before running this code, be sure for Matlab installation in your system.
#
# >data(Cow)
# >with(lrqmm_m(id=REGNO,sire=FREG,dam=MREG,X=HYS,Y=HL,cova=AGECAL,alpha=1,tau=0.5,
#  Factor=TRUE),data=Cow)
#
#
#End(not run)
# }