R Graphical Manual

Browse All

Last data update: 2014.03.03

R: GS-Estimates for multivariate regression with bootstrap...

FRBmultiregGS

R Documentation

GS-Estimates for multivariate regression with bootstrap confidence intervals

Description

Computes GS-estimates for multivariate regression together with standard errors, confidence intervals and p-values based on the Fast and Robust Bootstrap.

Usage

## S3 method for class 'formula'
FRBmultiregGS(formula, data=NULL, ...)

## Default S3 method:
FRBmultiregGS(X, Y, int = TRUE, R = 999, bdp = 0.5, conf = 0.95,
control=GScontrol(...), na.action=na.omit, ...)

Arguments

`formula`	an object of class `formula`; a symbolic description of the model to be fit.
`data`	data frame from which variables specified in formula are to be taken.
`X`	a matrix or data frame containing the explanatory variables.
`Y`	a matrix or data frame containing the response variables.
`int`	logical: if `TRUE` an intercept term is added to the model (unless it is already present in `X`)
`R`	number of bootstrap samples. Default is `R=999`.
`bdp`	required breakdown point. Should have 0 < `bdp` ≤ 0.5, the default is 0.5.
`conf`	confidence level of the bootstrap confidence intervals. Default is `conf=0.95`.
`control`	a list with control parameters for tuning the computing algorithm, see `GScontrol`().
`na.action`	a function which indicates what should happen when the data contain NAs. Defaults to `na.omit`.
`...`	allows for specifying control parameters directly instead of via `control`.

Details

Generalized S-estimators are defined by minimizing the determinant of a robust estimator of the scatter matrix of the differences of the residuals (Roelant et al. 2009). Hence, this procedure is intercept free and only gives an estimate for the slope matrix. To estimate the intercept, we use the M-type estimator of location of Lopuhaa (1992) on the residuals with the residual scatter matrix estimate of the residuals as a preliminary estimate. This computation is carried out by a call to GSest_multireg(), which uses a fast-S-type algorithm (its tuning parameters can be changed via the control argument). The result of this call is also returned as the value est.

The Fast and Robust Bootstrap (Salibian-Barrera and Zamar 2002) is used to calculate so-called basic bootstrap confidence intervals and bias corrected and accelerated (BCa) confidence intervals (Davison and Hinkley 1997, p.194 and p.204 respectively). Apart from the intervals with the requested confidence level, the function also returns p-values for each coefficient corresponding to the hypothesis that the actual coefficient is zero. The p-values are computed as 1 minus the smallest level for which the confidence intervals would include zero. Both BCa and basic bootstrap p-values in this sense are given. The bootstrap calculation is carried out by a call to GSboot_multireg(), the result of which is returned as the value bootest. Bootstrap standard errors are returned as well.

Note: Bootstrap samples which contain too few distinct observations with positive weights are discarded (a warning is given if this happens). The number of samples actually used is returned via ROK.

In the formula-interface, a multivariate response is produced via cbind. For example cbind(x4,x5) ~ x1+x2+x3. All arguments from the default method can also be passed to the formula method.

The returned object inherits from class mlm such that the standard coef, residuals, fitted and predict functions can be used.

Value

An object of class FRBmultireg which extends class mlm and contains at least the following components:

coefficients

GS-estimates of the regression coefficients

`residuals`	the residuals, that is response minus fitted values
`fitted.values`	the fitted values.
`Sigma`	GS-estimate of the error covariance matrix
`scale`	GS-estimate of the size of the multivariate errors
`weights`	implicit weights corresponding to the GS-estimates (i.e. final weights in the RWLS procedure for the intercept estimate)
`outFlag`	outlier flags: 1 if the robust distance of the residual exceeds the .975 quantile of (the square root of) the chi-square distribution with degrees of freedom equal to the dimension of the responses; 0 otherwise
`SE`	bootstrap standard errors corresponding to the regression coefficients
`cov`	bootstrap covariance matrix corresponding to the regression coefficients (in vectorized form)
`CI.bca.lower`	a matrix containing the lower bound of the bias corrected and accelerated confidence intervals for the regression coefficients
`CI.bca.upper`	a matrix containing the upper bound of the bias corrected and accelerated confidence intervals for the regression coefficients
`CI.basic.lower`	a matrix containing the lower bound of basic bootstrap intervals for the regression coefficients
`CI.basic.upper`	a matrix containing the upper bound of basic bootstrap intervals for the regression coefficients
`p.bca`	a matrix containing the p-values based on the BCa confidence intervals for the regression coefficients
`p.basic`	a matrix containing the p-values based on the basic bootstrap intervals for the regression coefficients
`est`	GS-estimates as returned by the call to `GSest_multireg`()
`bootest`	bootstrap results for the GS-estimates as returned by the call to `GSboot_multireg`()
`conf`	a copy of the `conf` argument
`method`	a list with following components: `est` = character string indicating that GS-estimates were used, and `bdp` = a copy of the `bdp` argument
`control`	a copy of the `control` argument
`X, Y`	either copies of the respective arguments or the corresponding matrices produced from `formula`
`ROK`	number of bootstrap samples actually used (i.e. not discarded due to too few distinct observations with positive weight)

Author(s)

Ella Roelant, Stefan Van Aelst and Gert Willems

References

A.C. Davison and D.V. Hinkley (1997) Bootstrap Methods and their Application. Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge: Cambridge University Press.
H.P. Lopuhaa (1992) Highly efficient estimators of multivariate location with high breakdown point. The Annals of Statistics, 20, 398-413.
E. Roelant, S. Van Aelst and C. Croux (2009) Multivariate Generalized S-estimators. Journal of Multivariate Analysis, 100, 876–887.
M. Salibian-Barrera, S. Van Aelst and G. Willems (2008) Fast and robust bootstrap. Statistical Methods and Applications, 17, 41-71.
S. Van Aelst and G. Willems (2013). Fast and robust bootstrap for multivariate inference: The R package FRB. Journal of Statistical Software, 53(3), 1–32. URL: http://www.jstatsoft.org/v53/i03/.

Examples

data(schooldata)
school.x <- data.matrix(schooldata[,1:5])
school.y <- data.matrix(schooldata[,6:8])

#computes 25% breakdown point GS-estimate and 80% confidence intervals 
#based on 99 bootstrap samples:
GSres <- FRBmultiregGS(school.x, school.y, R=99, bdp = 0.25, conf = 0.8,nsamp=50)
#or using the formula interface
## Not run: GSres <- FRBmultiregGS(cbind(reading,mathematics,selfesteem)~., data=schooldata, 
          bdp = 0.25, conf = 0.8,R=99)
## End(Not run)

#the print method just displays the coefficient estimates
GSres

#the summary function additionally displays the bootstrap standard errors and p-values 
#("BCA" method by default)
summary(GSres)

summary(GSres, confmethod="basic")

#ask explicitely for the coefficient matrix:
GSres$coefficients
# or equivalently,
coef(GSres)
#For the error covariance matrix:
GSres$Sigma
                                                              
#plot some bootstrap histograms for the coefficient estimates 
#(with "BCA" intervals by default) 
plot(GSres, expl=c("education", "occupation"), resp=c("selfesteem","reading"))

#plot bootstrap histograms for all coefficient estimates
plot(GSres)
#possibly the plot-function has made a selection of coefficients to plot here, 
#since 'all' may have been too many to fit on one page, see help(plot.FRBmultireg); 
#this is platform-dependent

Results


R version 3.3.1 (2016-06-21) -- "Bug in Your Hair"
Copyright (C) 2016 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

> library(FRB)
Loading required package: corpcor
Loading required package: rrcov
Loading required package: robustbase
Scalable Robust Estimators with High Breakdown Point (version 1.3-11)

> png(filename="/home/ddbj/snapshot/RGM3/R_CC/result/FRB/FRBmultiregGS.Rd_%03d_medium.png", width=480, height=480)
> ### Name: FRBmultiregGS
> ### Title: GS-Estimates for multivariate regression with bootstrap
> ###   confidence intervals
> ### Aliases: FRBmultiregGS FRBmultiregGS.default FRBmultiregGS.formula
> ### Keywords: multivariate robust
> 
> ### ** Examples
> 
> data(schooldata)
> school.x <- data.matrix(schooldata[,1:5])
> school.y <- data.matrix(schooldata[,6:8])
> 
> #computes 25% breakdown point GS-estimate and 80% confidence intervals 
> #based on 99 bootstrap samples:
> GSres <- FRBmultiregGS(school.x, school.y, R=99, bdp = 0.25, conf = 0.8,nsamp=50)
> #or using the formula interface
> ## Not run: 
> ##D GSres <- FRBmultiregGS(cbind(reading,mathematics,selfesteem)~., data=schooldata, 
> ##D           bdp = 0.25, conf = 0.8,R=99)
> ## End(Not run)
> 
> #the print method just displays the coefficient estimates
> GSres

Multivariate regression based on multivariate GS-estimates (breakdown point = 0.25) 

Coefficients:
            reading mathematics selfesteem
(intercept)  1.9406      2.4878     0.0754
education    0.1218      0.0124    -0.0467
occupation   4.9100      6.0603     2.1394
visit        0.0675      0.0197     0.2377
counseling  -0.7759     -0.8136    -0.0814
teacher     -0.1876     -0.2889     0.0228
> 
> #the summary function additionally displays the bootstrap standard errors and p-values 
> #("BCA" method by default)
> summary(GSres)

Multivariate regression based on GS-estimates (breakdown point = 0.25) 


Response reading:

Residuals:
    Min      1Q  Median      3Q     Max 
-14.697  -1.860   0.258   2.359  26.800 

Coefficients:
            Estimate   Std.Error   p-value    
(intercept)   1.9406      1.2097    0.0904   .
education     0.1218      1.2795    0.0000 ***
occupation    4.9100      0.3818    0.8898    
visit         0.0675      0.0727    0.0830   .
counseling   -0.7759      1.2053    0.0000 ***
teacher      -0.1876      0.3265    0.8331    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

p-values based on BCA method!


Response mathematics:

Residuals:
    Min      1Q  Median      3Q     Max 
-10.505  -2.837  -0.573   3.122  38.599 

Coefficients:
            Estimate   Std.Error   p-value    
(intercept)   2.4878       0.214    0.0000 ***
education     0.0124       0.153    0.1354    
occupation    6.0603       0.085    0.9282    
visit         0.0197       1.200    0.0000 ***
counseling   -0.8136       0.298    0.9596    
teacher      -0.2889       0.320    0.0451   *
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

p-values based on BCA method!


Response selfesteem:

Residuals:
    Min      1Q  Median      3Q     Max 
-2.6744 -0.6642  0.0195  0.6830  3.4961 

Coefficients:
            Estimate   Std.Error   p-value    
(intercept)   0.0754      0.1946     0.123    
education    -0.0467      0.0467     0.467    
occupation    2.1394      0.6071     0.000 ***
visit         0.2377      0.0920     0.000 ***
counseling   -0.0814      0.1330     0.462    
teacher       0.0228      0.0460     0.625    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

p-values based on BCA method!

Robust residual scale: 2.22 

Error covariance matrix estimate:
            reading mathematics selfesteem
reading        14.8       14.59       2.60
mathematics    14.6       21.59       2.65
selfesteem      2.6        2.65       1.59
> 
> summary(GSres, confmethod="basic")

Multivariate regression based on GS-estimates (breakdown point = 0.25) 


Response reading:

Residuals:
    Min      1Q  Median      3Q     Max 
-14.697  -1.860   0.258   2.359  26.800 

Coefficients:
            Estimate   Std.Error   p-value    
(intercept)   1.9406      1.2097     0.182    
education     0.1218      1.2795     0.101    
occupation    4.9100      0.3818     0.909    
visit         0.0675      0.0727     0.202    
counseling   -0.7759      1.2053     0.000 ***
teacher      -0.1876      0.3265     0.909    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

p-values based on Basic Bootstrap!


Response mathematics:

Residuals:
    Min      1Q  Median      3Q     Max 
-10.505  -2.837  -0.573   3.122  38.599 

Coefficients:
            Estimate   Std.Error   p-value    
(intercept)   2.4878       0.214     0.000 ***
education     0.0124       0.153     0.202    
occupation    6.0603       0.085     0.949    
visit         0.0197       1.200     0.000 ***
counseling   -0.8136       0.298     0.949    
teacher      -0.2889       0.320     0.000 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

p-values based on Basic Bootstrap!


Response selfesteem:

Residuals:
    Min      1Q  Median      3Q     Max 
-2.6744 -0.6642  0.0195  0.6830  3.4961 

Coefficients:
            Estimate   Std.Error   p-value    
(intercept)   0.0754      0.1946    0.2424    
education    -0.0467      0.0467    0.0808   .
occupation    2.1394      0.6071    0.0000 ***
visit         0.2377      0.0920    0.0000 ***
counseling   -0.0814      0.1330    0.4646    
teacher       0.0228      0.0460    0.4444    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

p-values based on Basic Bootstrap!

Robust residual scale: 2.22 

Error covariance matrix estimate:
            reading mathematics selfesteem
reading        14.8       14.59       2.60
mathematics    14.6       21.59       2.65
selfesteem      2.6        2.65       1.59
> 
> #ask explicitely for the coefficient matrix:
> GSres$coefficients
                reading mathematics  selfesteem
(intercept)  1.94063330  2.48784629  0.07544784
education    0.12180089  0.01239441 -0.04673212
occupation   4.90995552  6.06025859  2.13937389
visit        0.06746034  0.01970355  0.23773282
counseling  -0.77593115 -0.81356534 -0.08140681
teacher     -0.18758042 -0.28890086  0.02275939
> # or equivalently,
> coef(GSres)
                reading mathematics  selfesteem
(intercept)  1.94063330  2.48784629  0.07544784
education    0.12180089  0.01239441 -0.04673212
occupation   4.90995552  6.06025859  2.13937389
visit        0.06746034  0.01970355  0.23773282
counseling  -0.77593115 -0.81356534 -0.08140681
teacher     -0.18758042 -0.28890086  0.02275939
> #For the error covariance matrix:
> GSres$Sigma
              reading mathematics selfesteem
reading     14.775899   14.594412   2.598958
mathematics 14.594412   21.590800   2.647048
selfesteem   2.598958    2.647048   1.586931
>                                                               
> #plot some bootstrap histograms for the coefficient estimates 
> #(with "BCA" intervals by default) 
> plot(GSres, expl=c("education", "occupation"), resp=c("selfesteem","reading"))
> 
> #plot bootstrap histograms for all coefficient estimates
> plot(GSres)
Warning message:
In plot.FRBmultireg(GSres) :
  Number of plots too large to fit on the page, subset was selected: consider specifying (fewer) 
      variables in 'expl' and 'resp'; or enlarge graphics device; or set 'onepage=FALSE'
> #possibly the plot-function has made a selection of coefficients to plot here, 
> #since 'all' may have been too many to fit on one page, see help(plot.FRBmultireg); 
> #this is platform-dependent
> 
> 
> 
> 
> 
> 
> dev.off()
null device 
          1 
>