Nothing
R/measures.R
In rbvs: Ranking-Based Variable Selection
#' @title Measure an impact of the covariates on the response using Lasso
#' This function evaluates the Lasso coefficients regressing \code{y} onto the design matrix \code{x} over subsamples in \code{subsamples}.
#' @details To solve the Lasso problem, we implement the coordinate descent algorithm as in Breheny Jian (2011).
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param nonzero Number of non-zero coefficients estimated for each subsample.
#' @param family Determines the likelihood optimised in the estimation procedure.
#' @param alpha Scalar between 0 and 1 determining l2 penalty (see details).
#' @param maxit Maximum number of itarations when computing the lasso coefficients.
#' @param tol Scalar determining convergence of the lasso algorithm used.
#' @param lambda.ratio Scalar being a fraction of 1. Used in the lasso algorithm
#' @param nlam Number of penalty parameters used in the lasso algorithm.
#' @param ... Not in use.
#' @useDynLib rbvs lasso_coef_gaussian_r lasso_coef_binomial_r
#' @export
#' @author Rafal Baranowski, Patrick Breheny
#' @references Tibshirani, Robert. "Regression shrinkage and selection via the lasso." Journal of the Royal Statistical Society. Series B (Methodological) (1996): 267-288.
#' @references Breheny, Patrick, and Jian Huang. "Coordinate descent algorithms for nonconvex penalized regression, with applications to biological feature selection." The Annals of Applied Statistics 5.1 (2011): 232.
lasso.coef <- function(x,y,subsamples,
nonzero=NULL,
family=c("gaussian", "binomial"),
alpha=1.0,
maxit=500,
tol=1e-2,
lambda.ratio=1e-6,
nlam=25,
...){
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
alpha <- as.double(alpha)[1]
if(alpha < 0 || alpha > 1) stop("Parameter 'alpha' has to be between 0 and 1.")
maxit <- as.integer(maxit)[1]
if(maxit < 50){
warning("Paramater 'maxit' must be greater than or equal to 50. Setting 'maxit' to 50.")
maxit <- 50
}
tol <- as.double(tol)[1]
if(tol <= 0) stop("'tol' has to be positive.")
lambda.ratio <- as.double(lambda.ratio)[1]
if(lambda.ratio < 0 || lambda.ratio > 1) stop("'lambda.ratio' has to be between 0 and 1.")
nlam <- as.integer(nlam)[1]
if(nlam < 2) {
warning("Paramater 'nlam' must be greater than or equal to 50. Setting 'nlam' to 2.")
maxit <- 2
}
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
m <- nrow(subsamples);
B <- ncol(subsamples);
p <- ncol(x)
n <- nrow(x)
if(is.null(nonzero)) nonzero <- m-1;
family <- match.arg(family)
if( family=="gaussian"){
return(.Call("lasso_coef_gaussian_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}else{
return(.Call("lasso_coef_binomial_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}
}
#' @title Measure an impact of the covariates on the response using MC+.
#' This function evaluates the MC+ coefficients regressing \code{y} onto the design matrix \code{x} over subsamples in \code{subsamples}.
#' @details To solve the MC+ problem, we implement the coordinate descent algorithm as in Breheny Jian (2011).
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param nonzero Number of non-zero coefficients estimated for each subsample.
#' @param family Determines the likelihood optimised in the estimation procedure.
#' @param alpha Scalar between 0 and 1 determining l2 penalty (see details).
#' @param gamma Scalar greater than 1. The concacivity parameter (see details).
#' @param maxit Maximum number of itarations when computing the MC+ coefficients.
#' @param tol Scalar determining convergence of the MC+ algorithm used.
#' @param lambda.ratio Scalar being a fraction of 1. Used in the MC+ algorithm
#' @param nlam Number of penalty parameters used in the MC+ algorithm.
#' @param ... Not in use.
#' @useDynLib rbvs mcplus_coef_gaussian_r mcplus_coef_binomial_r
#' @author Rafal Baranowski, Patrick Breheny
#' @export
#' @references Zhang, Cun-Hui. "Nearly unbiased variable selection under minimax concave penalty." The Annals of Statistics (2010): 894-942.
#' @references Breheny, Patrick, and Jian Huang. "Coordinate descent algorithms for nonconvex penalized regression, with applications to biological feature selection." The Annals of Applied Statistics 5.1 (2011): 232.
mcplus.coef <- function(x,y,subsamples,
nonzero=NULL,
family=c("gaussian", "binomial"),
alpha=1.0,
gamma=3.0,
maxit=500,
tol=1e-2,
lambda.ratio=1e-6,
nlam=25,
...){
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
alpha <- as.double(alpha)[1]
if(alpha < 0 || alpha > 1) stop("Parameter 'alpha' has to be between 0 and 1.")
gamma <- as.double(gamma)[1]
if(gamma < 1) stop("'gamma' must be greater than 1.")
maxit <- as.integer(maxit)[1]
if(maxit < 50){
warning("Paramater 'maxit' must be greater than or equal to 50. Setting 'maxit' to 50.")
maxit <- 50
}
tol <- as.double(tol)[1]
if(tol <= 0) stop("'tol' has to be positive.")
lambda.ratio <- as.double(lambda.ratio)[1]
if(lambda.ratio < 0 || lambda.ratio > 1) stop("'lambda.ratio' has to be between 0 and 1.")
nlam <- as.integer(nlam)[1]
if(nlam < 2) {
warning("Paramater 'nlam' must be greater than or equal to 50. Setting 'nlam' to 2.")
maxit <- 2
}
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
m <- nrow(subsamples);
B <- ncol(subsamples);
p <- ncol(x)
n <- nrow(x)
if(is.null(nonzero)) nonzero <- m-1;
family <- match.arg(family)
if( family=="gaussian"){
return(.Call("mcplus_coef_gaussian_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(gamma),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}else{
return(.Call("mcplus_coef_binomial_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(gamma),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}
}
#' @title Measure an impact of the covariates on the response using Pearson correlatio.
#' This function evaluates the Pearson correlation coefficient between the response \code{y} and each column in the design matrix \code{x} over subsamples in \code{subsamples}.
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param ... Not in use.
#' @return Numeric \code{p} by \code{N} matrix with Pearson correlations evaluated for each subsample.
#' @useDynLib rbvs pearson_cor_r
#' @export
pearson.cor <- function(x,y,subsamples, ...){
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
return(.Call("pearson_cor_r",
subsamples,
x,
y))
}
#' @title Measure an impact of the covariates on the response using the distance correlation
#' This function evaluates the distance correlation between the response \code{y} and each column in the design matrix \code{x} over subsamples in \code{subsamples}.
#' @useDynLib rbvs distance_cor_r
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param index Positive scalar.
#' @param ... Not in use.
#' @return Numeric \code{p} by \code{N} matrix with distance correlations evaluated for each subsample.
#' @references Maria L. Rizzo and Gabor J. Szekely (2014). energy: E-statistics
#' (energy statistics). R package version 1.6.1 (\url{http://CRAN.R-project.org/package=energy}).
#' @export
distance.cor <- function(x,y, subsamples, index=1, ...){
#guarantee data types correct for C function
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
index <- as.numeric(index)
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
if(any(is.na(index))) stop("Scalar 'index' cannot contain NA's.")
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
if(length(index) > 1){
warning("Variable 'index' contains too many elements. Taking just the first one.")
index <- index[1]
}else if (length(index) <1) warning("Variable 'index' contains too few elements. Setting 'index' to 1.")
if(index < 0) stop("Variable 'index' must be positive.")
# Call C function calculating correlations
return(.Call("distance_cor_r",
subsamples,
x,
y,
as.numeric(index)))
}
Try the rbvs package in your browser
Any scripts or data that you put into this service are public.
rbvs documentation built on May 2, 2019, 7:31 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
#' @title Measure an impact of the covariates on the response using Lasso
#' This function evaluates the Lasso coefficients regressing \code{y} onto the design matrix \code{x} over subsamples in \code{subsamples}.
#' @details To solve the Lasso problem, we implement the coordinate descent algorithm as in Breheny Jian (2011).
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param nonzero Number of non-zero coefficients estimated for each subsample.
#' @param family Determines the likelihood optimised in the estimation procedure.
#' @param alpha Scalar between 0 and 1 determining l2 penalty (see details).
#' @param maxit Maximum number of itarations when computing the lasso coefficients.
#' @param tol Scalar determining convergence of the lasso algorithm used.
#' @param lambda.ratio Scalar being a fraction of 1. Used in the lasso algorithm
#' @param nlam Number of penalty parameters used in the lasso algorithm.
#' @param ... Not in use.
#' @useDynLib rbvs lasso_coef_gaussian_r lasso_coef_binomial_r
#' @export
#' @author Rafal Baranowski, Patrick Breheny
#' @references Tibshirani, Robert. "Regression shrinkage and selection via the lasso." Journal of the Royal Statistical Society. Series B (Methodological) (1996): 267-288.
#' @references Breheny, Patrick, and Jian Huang. "Coordinate descent algorithms for nonconvex penalized regression, with applications to biological feature selection." The Annals of Applied Statistics 5.1 (2011): 232.
lasso.coef <- function(x,y,subsamples,
nonzero=NULL,
family=c("gaussian", "binomial"),
alpha=1.0,
maxit=500,
tol=1e-2,
lambda.ratio=1e-6,
nlam=25,
...){
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
alpha <- as.double(alpha)[1]
if(alpha < 0 || alpha > 1) stop("Parameter 'alpha' has to be between 0 and 1.")
maxit <- as.integer(maxit)[1]
if(maxit < 50){
warning("Paramater 'maxit' must be greater than or equal to 50. Setting 'maxit' to 50.")
maxit <- 50
}
tol <- as.double(tol)[1]
if(tol <= 0) stop("'tol' has to be positive.")
lambda.ratio <- as.double(lambda.ratio)[1]
if(lambda.ratio < 0 || lambda.ratio > 1) stop("'lambda.ratio' has to be between 0 and 1.")
nlam <- as.integer(nlam)[1]
if(nlam < 2) {
warning("Paramater 'nlam' must be greater than or equal to 50. Setting 'nlam' to 2.")
maxit <- 2
}
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
m <- nrow(subsamples);
B <- ncol(subsamples);
p <- ncol(x)
n <- nrow(x)
if(is.null(nonzero)) nonzero <- m-1;
family <- match.arg(family)
if( family=="gaussian"){
return(.Call("lasso_coef_gaussian_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}else{
return(.Call("lasso_coef_binomial_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}
}
#' @title Measure an impact of the covariates on the response using MC+.
#' This function evaluates the MC+ coefficients regressing \code{y} onto the design matrix \code{x} over subsamples in \code{subsamples}.
#' @details To solve the MC+ problem, we implement the coordinate descent algorithm as in Breheny Jian (2011).
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param nonzero Number of non-zero coefficients estimated for each subsample.
#' @param family Determines the likelihood optimised in the estimation procedure.
#' @param alpha Scalar between 0 and 1 determining l2 penalty (see details).
#' @param gamma Scalar greater than 1. The concacivity parameter (see details).
#' @param maxit Maximum number of itarations when computing the MC+ coefficients.
#' @param tol Scalar determining convergence of the MC+ algorithm used.
#' @param lambda.ratio Scalar being a fraction of 1. Used in the MC+ algorithm
#' @param nlam Number of penalty parameters used in the MC+ algorithm.
#' @param ... Not in use.
#' @useDynLib rbvs mcplus_coef_gaussian_r mcplus_coef_binomial_r
#' @author Rafal Baranowski, Patrick Breheny
#' @export
#' @references Zhang, Cun-Hui. "Nearly unbiased variable selection under minimax concave penalty." The Annals of Statistics (2010): 894-942.
#' @references Breheny, Patrick, and Jian Huang. "Coordinate descent algorithms for nonconvex penalized regression, with applications to biological feature selection." The Annals of Applied Statistics 5.1 (2011): 232.
mcplus.coef <- function(x,y,subsamples,
nonzero=NULL,
family=c("gaussian", "binomial"),
alpha=1.0,
gamma=3.0,
maxit=500,
tol=1e-2,
lambda.ratio=1e-6,
nlam=25,
...){
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
alpha <- as.double(alpha)[1]
if(alpha < 0 || alpha > 1) stop("Parameter 'alpha' has to be between 0 and 1.")
gamma <- as.double(gamma)[1]
if(gamma < 1) stop("'gamma' must be greater than 1.")
maxit <- as.integer(maxit)[1]
if(maxit < 50){
warning("Paramater 'maxit' must be greater than or equal to 50. Setting 'maxit' to 50.")
maxit <- 50
}
tol <- as.double(tol)[1]
if(tol <= 0) stop("'tol' has to be positive.")
lambda.ratio <- as.double(lambda.ratio)[1]
if(lambda.ratio < 0 || lambda.ratio > 1) stop("'lambda.ratio' has to be between 0 and 1.")
nlam <- as.integer(nlam)[1]
if(nlam < 2) {
warning("Paramater 'nlam' must be greater than or equal to 50. Setting 'nlam' to 2.")
maxit <- 2
}
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
m <- nrow(subsamples);
B <- ncol(subsamples);
p <- ncol(x)
n <- nrow(x)
if(is.null(nonzero)) nonzero <- m-1;
family <- match.arg(family)
if( family=="gaussian"){
return(.Call("mcplus_coef_gaussian_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(gamma),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}else{
return(.Call("mcplus_coef_binomial_r",
as.matrix(subsamples),
as.matrix(x),
as.numeric(y),
as.integer(nonzero),
as.double(alpha),
as.double(gamma),
as.double(tol),
as.integer(maxit),
as.double(lambda.ratio),
as.integer(nlam)))
}
}
#' @title Measure an impact of the covariates on the response using Pearson correlatio.
#' This function evaluates the Pearson correlation coefficient between the response \code{y} and each column in the design matrix \code{x} over subsamples in \code{subsamples}.
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param ... Not in use.
#' @return Numeric \code{p} by \code{N} matrix with Pearson correlations evaluated for each subsample.
#' @useDynLib rbvs pearson_cor_r
#' @export
pearson.cor <- function(x,y,subsamples, ...){
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
return(.Call("pearson_cor_r",
subsamples,
x,
y))
}
#' @title Measure an impact of the covariates on the response using the distance correlation
#' This function evaluates the distance correlation between the response \code{y} and each column in the design matrix \code{x} over subsamples in \code{subsamples}.
#' @useDynLib rbvs distance_cor_r
#' @param x Matrix with \code{n} observations of \code{p} covariates in each row.
#' @param y Response vector with \code{n} observations.
#' @param subsamples Matrix with \code{m} indices of \code{N} subsamples in each column.
#' @param index Positive scalar.
#' @param ... Not in use.
#' @return Numeric \code{p} by \code{N} matrix with distance correlations evaluated for each subsample.
#' @references Maria L. Rizzo and Gabor J. Szekely (2014). energy: E-statistics
#' (energy statistics). R package version 1.6.1 (\url{http://CRAN.R-project.org/package=energy}).
#' @export
distance.cor <- function(x,y, subsamples, index=1, ...){
#guarantee data types correct for C function
x <- as.matrix(x)
storage.mode(x) <- "double"
y <- as.numeric(y)
storage.mode(y) <- "double"
subsamples <- as.matrix(subsamples)
storage.mode(subsamples) <- "integer"
index <- as.numeric(index)
#check if the sizes of x and y match
n <- length(y)
if(nrow(x) != n) stop("The number of rows in 'x' must be equal to the length of 'y'.")
#check for NA's
if(any(is.na(x))) stop("Matix 'x' cannot contain NA's.")
if(any(is.na(y))) stop("Vector 'y' cannot contain NA's.")
if(any(is.na(subsamples))) stop("Matrix 'subsamples' cannot contain NA's.")
if(any(is.na(index))) stop("Scalar 'index' cannot contain NA's.")
#check if subsamples are actually subsamples
min.ind <- min(subsamples)
max.ind <- max(subsamples)
if(min.ind < 1 || max.ind > n) stop("Elements of 'subsamples' must be between 1 and length(y).")
if(length(index) > 1){
warning("Variable 'index' contains too many elements. Taking just the first one.")
index <- index[1]
}else if (length(index) <1) warning("Variable 'index' contains too few elements. Setting 'index' to 1.")
if(index < 0) stop("Variable 'index' must be positive.")
# Call C function calculating correlations
return(.Call("distance_cor_r",
subsamples,
x,
y,
as.numeric(index)))
}
Try the rbvs package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.