CRAN Package Check Results for Package probsvm

Last updated on 2020-02-19 10:49:05 CET.

Flavor Version Tinstall Tcheck Ttotal Status Flags
r-devel-linux-x86_64-debian-clang 1.00 4.95 43.58 48.53 ERROR
r-devel-linux-x86_64-debian-gcc 1.00 4.29 34.46 38.75 ERROR
r-devel-linux-x86_64-fedora-clang 1.00 59.56 ERROR
r-devel-linux-x86_64-fedora-gcc 1.00 57.49 ERROR
r-devel-windows-ix86+x86_64 1.00 11.00 52.00 63.00 OK
r-devel-windows-ix86+x86_64-gcc8 1.00 13.00 53.00 66.00 OK
r-patched-linux-x86_64 1.00 3.57 38.94 42.51 OK
r-patched-solaris-x86 1.00 76.00 OK
r-release-linux-x86_64 1.00 4.21 39.02 43.23 OK
r-release-windows-ix86+x86_64 1.00 10.00 47.00 57.00 OK
r-release-osx-x86_64 1.00 OK
r-oldrel-windows-ix86+x86_64 1.00 8.00 48.00 56.00 OK
r-oldrel-osx-x86_64 1.00 OK

Check Details

Version: 1.00
Check: examples
Result: ERROR
    Running examples in 'probsvm-Ex.R' failed
    The error most likely occurred in:
    
    > base::assign(".ptime", proc.time(), pos = "CheckExEnv")
    > ### Name: probsvm
    > ### Title: Main function that provides models for multiclass conditional
    > ### probability estimation and label prediction
    > ### Aliases: probsvm
    >
    > ### ** Examples
    >
    > # iris data #
    >
    > data(iris)
    >
    > iris.x=iris[c(1:20,51:70,101:120),-5]
    >
    > iris.y=iris[c(1:20,51:70,101:120),5]
    >
    > iris.test=iris[c(21:50,71:100,121:150),-5]
    >
    > a = probsvm(iris.x,iris.y,type="ovo",
    + Inum=10,fold=2,lambdas=2^seq(-10,10,by=3))
     ----------- FAILURE REPORT --------------
     --- failure: the condition has length > 1 ---
     --- srcref ---
    :
     --- package (from environment) ---
    probsvm
     --- call from context ---
    predict_pipathresults(obj, x.test, pi = I)
     --- call from argument ---
    if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
    }
     --- R stacktrace ---
    where 1: predict_pipathresults(obj, x.test, pi = I)
    where 2: prob.svm(x.train, y.train, x.test, lambda = lambda, Inum = Inum,
     kernel = kernel, kparam = kparam)
    where 3: loglik.svm(x.train.temp, y.train.temp, x.tune.temp, y.tune.temp,
     lambda = lambda, Inum = Inum, kernel = kernel, kparam = kparam)
    where 4: probsvm(iris.x, iris.y, type = "ovo", Inum = 10, fold = 2, lambdas = 2^seq(-10,
     10, by = 3))
    
     --- value of length: 2 type: logical ---
    [1] FALSE TRUE
     --- function from context ---
    function (obj, new.x = NULL, pi = NULL)
    {
     obj.pi = obj$pi[1]
     obj.alpha0 = obj$alpha0[1]
     obj.alpha = obj$alpha[1, ]
     for (ii in 2:length(obj$pi)) {
     if (obj$pi[ii] != obj$pi[ii - 1]) {
     obj.pi = c(obj.pi, obj$pi[ii])
     obj.alpha0 = c(obj.alpha0, obj$alpha0[ii])
     obj.alpha = rbind(obj.alpha, obj$alpha[ii, ])
     }
     }
     kernel = obj$kernel
     kparam = obj$kparam
     if (is.null(new.x)) {
     new.x = obj$x
     }
     if (is.null(pi)) {
     pi = obj.pi
     }
     if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
     }
     if (class(new.x) == "data.frame") {
     new.x = as.matrix(new.x)
     }
     if (ncol(new.x) != ncol(obj$x)) {
     stop("The new covariates matrix has a wrong dimension.")
     }
     if (!is.numeric(pi)) {
     stop("The parameter pi must be numeric.")
     }
     if (min(pi) < 0 | max(pi) > 1) {
     stop("The parameter pi must be in [0,1].")
     }
     K <- Kmat(new.x, obj$x, kernel, kparam)
     pred.y = numeric(0)
     alpha0 = numeric(0)
     alpha = numeric(0)
     f.hat = numeric(0)
     for (i in 1:length(pi)) {
     temp = pi[i]
     index = which(obj.pi == temp)
     if (length(index) == 1) {
     temp.alpha = obj.alpha[index, ]
     temp.alpha0 = obj.alpha0[index]
     new.y1 = K %*% temp.alpha + temp.alpha0
     }
     if (length(index) == 0) {
     if (temp < (obj.pi[1])) {
     temp.alpha = obj.alpha[1, ]
     temp.alpha0 = obj.alpha0[1]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp > (obj.pi[length(obj.pi)])) {
     temp.alpha = obj.alpha[length(obj.pi), ]
     temp.alpha0 = obj.alpha0[length(obj.pi)]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp < (obj.pi[length(obj.pi)]) & temp > (obj.pi[1])) {
     index2 = max(which(obj.pi < temp))
     temp.alpha = obj.alpha[index2, ] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha[(index2 +
     1), ] - obj.alpha[index2, ])
     temp.alpha0 = obj.alpha0[index2] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha0[(index2 +
     1)] - obj.alpha0[index2])
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     }
     f.hat = cbind(f.hat, new.y1)
     new.y1 = as.numeric(new.y1 > 0) * 2 - 1
     pred.y = cbind(pred.y, new.y1)
     alpha = cbind(alpha, temp.alpha)
     alpha0 = c(alpha0, temp.alpha0)
     }
     colnames(f.hat) = NULL
     colnames(alpha) = NULL
     colnames(pred.y) = NULL
     z = list(pi = obj.pi, fitted.alpha0 = alpha0, fitted.alpha = alpha,
     fitted.f = f.hat, predicted.y = pred.y)
     return(z)
    }
    <bytecode: 0x320b9e0>
    <environment: namespace:probsvm>
     --- function search by body ---
    Function predict_pipathresults in namespace probsvm has this body.
     ----------- END OF FAILURE REPORT --------------
    Error in if (class(new.x) != "matrix" & class(new.x) != "data.frame") { :
     the condition has length > 1
    Calls: probsvm -> loglik.svm -> prob.svm -> predict_pipathresults
    Execution halted
Flavor: r-devel-linux-x86_64-debian-clang

Version: 1.00
Check: examples
Result: ERROR
    Running examples in ‘probsvm-Ex.R’ failed
    The error most likely occurred in:
    
    > base::assign(".ptime", proc.time(), pos = "CheckExEnv")
    > ### Name: probsvm
    > ### Title: Main function that provides models for multiclass conditional
    > ### probability estimation and label prediction
    > ### Aliases: probsvm
    >
    > ### ** Examples
    >
    > # iris data #
    >
    > data(iris)
    >
    > iris.x=iris[c(1:20,51:70,101:120),-5]
    >
    > iris.y=iris[c(1:20,51:70,101:120),5]
    >
    > iris.test=iris[c(21:50,71:100,121:150),-5]
    >
    > a = probsvm(iris.x,iris.y,type="ovo",
    + Inum=10,fold=2,lambdas=2^seq(-10,10,by=3))
     ----------- FAILURE REPORT --------------
     --- failure: the condition has length > 1 ---
     --- srcref ---
    :
     --- package (from environment) ---
    probsvm
     --- call from context ---
    predict_pipathresults(obj, x.test, pi = I)
     --- call from argument ---
    if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
    }
     --- R stacktrace ---
    where 1: predict_pipathresults(obj, x.test, pi = I)
    where 2: prob.svm(x.train, y.train, x.test, lambda = lambda, Inum = Inum,
     kernel = kernel, kparam = kparam)
    where 3: loglik.svm(x.train.temp, y.train.temp, x.tune.temp, y.tune.temp,
     lambda = lambda, Inum = Inum, kernel = kernel, kparam = kparam)
    where 4: probsvm(iris.x, iris.y, type = "ovo", Inum = 10, fold = 2, lambdas = 2^seq(-10,
     10, by = 3))
    
     --- value of length: 2 type: logical ---
    [1] FALSE TRUE
     --- function from context ---
    function (obj, new.x = NULL, pi = NULL)
    {
     obj.pi = obj$pi[1]
     obj.alpha0 = obj$alpha0[1]
     obj.alpha = obj$alpha[1, ]
     for (ii in 2:length(obj$pi)) {
     if (obj$pi[ii] != obj$pi[ii - 1]) {
     obj.pi = c(obj.pi, obj$pi[ii])
     obj.alpha0 = c(obj.alpha0, obj$alpha0[ii])
     obj.alpha = rbind(obj.alpha, obj$alpha[ii, ])
     }
     }
     kernel = obj$kernel
     kparam = obj$kparam
     if (is.null(new.x)) {
     new.x = obj$x
     }
     if (is.null(pi)) {
     pi = obj.pi
     }
     if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
     }
     if (class(new.x) == "data.frame") {
     new.x = as.matrix(new.x)
     }
     if (ncol(new.x) != ncol(obj$x)) {
     stop("The new covariates matrix has a wrong dimension.")
     }
     if (!is.numeric(pi)) {
     stop("The parameter pi must be numeric.")
     }
     if (min(pi) < 0 | max(pi) > 1) {
     stop("The parameter pi must be in [0,1].")
     }
     K <- Kmat(new.x, obj$x, kernel, kparam)
     pred.y = numeric(0)
     alpha0 = numeric(0)
     alpha = numeric(0)
     f.hat = numeric(0)
     for (i in 1:length(pi)) {
     temp = pi[i]
     index = which(obj.pi == temp)
     if (length(index) == 1) {
     temp.alpha = obj.alpha[index, ]
     temp.alpha0 = obj.alpha0[index]
     new.y1 = K %*% temp.alpha + temp.alpha0
     }
     if (length(index) == 0) {
     if (temp < (obj.pi[1])) {
     temp.alpha = obj.alpha[1, ]
     temp.alpha0 = obj.alpha0[1]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp > (obj.pi[length(obj.pi)])) {
     temp.alpha = obj.alpha[length(obj.pi), ]
     temp.alpha0 = obj.alpha0[length(obj.pi)]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp < (obj.pi[length(obj.pi)]) & temp > (obj.pi[1])) {
     index2 = max(which(obj.pi < temp))
     temp.alpha = obj.alpha[index2, ] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha[(index2 +
     1), ] - obj.alpha[index2, ])
     temp.alpha0 = obj.alpha0[index2] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha0[(index2 +
     1)] - obj.alpha0[index2])
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     }
     f.hat = cbind(f.hat, new.y1)
     new.y1 = as.numeric(new.y1 > 0) * 2 - 1
     pred.y = cbind(pred.y, new.y1)
     alpha = cbind(alpha, temp.alpha)
     alpha0 = c(alpha0, temp.alpha0)
     }
     colnames(f.hat) = NULL
     colnames(alpha) = NULL
     colnames(pred.y) = NULL
     z = list(pi = obj.pi, fitted.alpha0 = alpha0, fitted.alpha = alpha,
     fitted.f = f.hat, predicted.y = pred.y)
     return(z)
    }
    <bytecode: 0x55592fb211a0>
    <environment: namespace:probsvm>
     --- function search by body ---
    Function predict_pipathresults in namespace probsvm has this body.
     ----------- END OF FAILURE REPORT --------------
    Error in if (class(new.x) != "matrix" & class(new.x) != "data.frame") { :
     the condition has length > 1
    Calls: probsvm -> loglik.svm -> prob.svm -> predict_pipathresults
    Execution halted
Flavor: r-devel-linux-x86_64-debian-gcc

Version: 1.00
Check: examples
Result: ERROR
    Running examples in ‘probsvm-Ex.R’ failed
    The error most likely occurred in:
    
    > ### Name: probsvm
    > ### Title: Main function that provides models for multiclass conditional
    > ### probability estimation and label prediction
    > ### Aliases: probsvm
    >
    > ### ** Examples
    >
    > # iris data #
    >
    > data(iris)
    >
    > iris.x=iris[c(1:20,51:70,101:120),-5]
    >
    > iris.y=iris[c(1:20,51:70,101:120),5]
    >
    > iris.test=iris[c(21:50,71:100,121:150),-5]
    >
    > a = probsvm(iris.x,iris.y,type="ovo",
    + Inum=10,fold=2,lambdas=2^seq(-10,10,by=3))
     ----------- FAILURE REPORT --------------
     --- failure: the condition has length > 1 ---
     --- srcref ---
    :
     --- package (from environment) ---
    probsvm
     --- call from context ---
    predict_pipathresults(obj, x.test, pi = I)
     --- call from argument ---
    if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
    }
     --- R stacktrace ---
    where 1: predict_pipathresults(obj, x.test, pi = I)
    where 2: prob.svm(x.train, y.train, x.test, lambda = lambda, Inum = Inum,
     kernel = kernel, kparam = kparam)
    where 3: loglik.svm(x.train.temp, y.train.temp, x.tune.temp, y.tune.temp,
     lambda = lambda, Inum = Inum, kernel = kernel, kparam = kparam)
    where 4: probsvm(iris.x, iris.y, type = "ovo", Inum = 10, fold = 2, lambdas = 2^seq(-10,
     10, by = 3))
    
     --- value of length: 2 type: logical ---
    [1] FALSE TRUE
     --- function from context ---
    function (obj, new.x = NULL, pi = NULL)
    {
     obj.pi = obj$pi[1]
     obj.alpha0 = obj$alpha0[1]
     obj.alpha = obj$alpha[1, ]
     for (ii in 2:length(obj$pi)) {
     if (obj$pi[ii] != obj$pi[ii - 1]) {
     obj.pi = c(obj.pi, obj$pi[ii])
     obj.alpha0 = c(obj.alpha0, obj$alpha0[ii])
     obj.alpha = rbind(obj.alpha, obj$alpha[ii, ])
     }
     }
     kernel = obj$kernel
     kparam = obj$kparam
     if (is.null(new.x)) {
     new.x = obj$x
     }
     if (is.null(pi)) {
     pi = obj.pi
     }
     if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
     }
     if (class(new.x) == "data.frame") {
     new.x = as.matrix(new.x)
     }
     if (ncol(new.x) != ncol(obj$x)) {
     stop("The new covariates matrix has a wrong dimension.")
     }
     if (!is.numeric(pi)) {
     stop("The parameter pi must be numeric.")
     }
     if (min(pi) < 0 | max(pi) > 1) {
     stop("The parameter pi must be in [0,1].")
     }
     K <- Kmat(new.x, obj$x, kernel, kparam)
     pred.y = numeric(0)
     alpha0 = numeric(0)
     alpha = numeric(0)
     f.hat = numeric(0)
     for (i in 1:length(pi)) {
     temp = pi[i]
     index = which(obj.pi == temp)
     if (length(index) == 1) {
     temp.alpha = obj.alpha[index, ]
     temp.alpha0 = obj.alpha0[index]
     new.y1 = K %*% temp.alpha + temp.alpha0
     }
     if (length(index) == 0) {
     if (temp < (obj.pi[1])) {
     temp.alpha = obj.alpha[1, ]
     temp.alpha0 = obj.alpha0[1]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp > (obj.pi[length(obj.pi)])) {
     temp.alpha = obj.alpha[length(obj.pi), ]
     temp.alpha0 = obj.alpha0[length(obj.pi)]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp < (obj.pi[length(obj.pi)]) & temp > (obj.pi[1])) {
     index2 = max(which(obj.pi < temp))
     temp.alpha = obj.alpha[index2, ] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha[(index2 +
     1), ] - obj.alpha[index2, ])
     temp.alpha0 = obj.alpha0[index2] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha0[(index2 +
     1)] - obj.alpha0[index2])
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     }
     f.hat = cbind(f.hat, new.y1)
     new.y1 = as.numeric(new.y1 > 0) * 2 - 1
     pred.y = cbind(pred.y, new.y1)
     alpha = cbind(alpha, temp.alpha)
     alpha0 = c(alpha0, temp.alpha0)
     }
     colnames(f.hat) = NULL
     colnames(alpha) = NULL
     colnames(pred.y) = NULL
     z = list(pi = obj.pi, fitted.alpha0 = alpha0, fitted.alpha = alpha,
     fitted.f = f.hat, predicted.y = pred.y)
     return(z)
    }
    <bytecode: 0x42f46d0>
    <environment: namespace:probsvm>
     --- function search by body ---
    Function predict_pipathresults in namespace probsvm has this body.
     ----------- END OF FAILURE REPORT --------------
    Error in if (class(new.x) != "matrix" & class(new.x) != "data.frame") { :
     the condition has length > 1
    Calls: probsvm -> loglik.svm -> prob.svm -> predict_pipathresults
    Execution halted
Flavor: r-devel-linux-x86_64-fedora-clang

Version: 1.00
Check: examples
Result: ERROR
    Running examples in ‘probsvm-Ex.R’ failed
    The error most likely occurred in:
    
    > ### Name: probsvm
    > ### Title: Main function that provides models for multiclass conditional
    > ### probability estimation and label prediction
    > ### Aliases: probsvm
    >
    > ### ** Examples
    >
    > # iris data #
    >
    > data(iris)
    >
    > iris.x=iris[c(1:20,51:70,101:120),-5]
    >
    > iris.y=iris[c(1:20,51:70,101:120),5]
    >
    > iris.test=iris[c(21:50,71:100,121:150),-5]
    >
    > a = probsvm(iris.x,iris.y,type="ovo",
    + Inum=10,fold=2,lambdas=2^seq(-10,10,by=3))
     ----------- FAILURE REPORT --------------
     --- failure: the condition has length > 1 ---
     --- srcref ---
    :
     --- package (from environment) ---
    probsvm
     --- call from context ---
    predict_pipathresults(obj, x.test, pi = I)
     --- call from argument ---
    if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
    }
     --- R stacktrace ---
    where 1: predict_pipathresults(obj, x.test, pi = I)
    where 2: prob.svm(x.train, y.train, x.test, lambda = lambda, Inum = Inum,
     kernel = kernel, kparam = kparam)
    where 3: loglik.svm(x.train.temp, y.train.temp, x.tune.temp, y.tune.temp,
     lambda = lambda, Inum = Inum, kernel = kernel, kparam = kparam)
    where 4: probsvm(iris.x, iris.y, type = "ovo", Inum = 10, fold = 2, lambdas = 2^seq(-10,
     10, by = 3))
    
     --- value of length: 2 type: logical ---
    [1] FALSE TRUE
     --- function from context ---
    function (obj, new.x = NULL, pi = NULL)
    {
     obj.pi = obj$pi[1]
     obj.alpha0 = obj$alpha0[1]
     obj.alpha = obj$alpha[1, ]
     for (ii in 2:length(obj$pi)) {
     if (obj$pi[ii] != obj$pi[ii - 1]) {
     obj.pi = c(obj.pi, obj$pi[ii])
     obj.alpha0 = c(obj.alpha0, obj$alpha0[ii])
     obj.alpha = rbind(obj.alpha, obj$alpha[ii, ])
     }
     }
     kernel = obj$kernel
     kparam = obj$kparam
     if (is.null(new.x)) {
     new.x = obj$x
     }
     if (is.null(pi)) {
     pi = obj.pi
     }
     if (class(new.x) != "matrix" & class(new.x) != "data.frame") {
     stop("The new covariates must be either a matrix or a data.frame.")
     }
     if (class(new.x) == "data.frame") {
     new.x = as.matrix(new.x)
     }
     if (ncol(new.x) != ncol(obj$x)) {
     stop("The new covariates matrix has a wrong dimension.")
     }
     if (!is.numeric(pi)) {
     stop("The parameter pi must be numeric.")
     }
     if (min(pi) < 0 | max(pi) > 1) {
     stop("The parameter pi must be in [0,1].")
     }
     K <- Kmat(new.x, obj$x, kernel, kparam)
     pred.y = numeric(0)
     alpha0 = numeric(0)
     alpha = numeric(0)
     f.hat = numeric(0)
     for (i in 1:length(pi)) {
     temp = pi[i]
     index = which(obj.pi == temp)
     if (length(index) == 1) {
     temp.alpha = obj.alpha[index, ]
     temp.alpha0 = obj.alpha0[index]
     new.y1 = K %*% temp.alpha + temp.alpha0
     }
     if (length(index) == 0) {
     if (temp < (obj.pi[1])) {
     temp.alpha = obj.alpha[1, ]
     temp.alpha0 = obj.alpha0[1]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp > (obj.pi[length(obj.pi)])) {
     temp.alpha = obj.alpha[length(obj.pi), ]
     temp.alpha0 = obj.alpha0[length(obj.pi)]
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     if (temp < (obj.pi[length(obj.pi)]) & temp > (obj.pi[1])) {
     index2 = max(which(obj.pi < temp))
     temp.alpha = obj.alpha[index2, ] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha[(index2 +
     1), ] - obj.alpha[index2, ])
     temp.alpha0 = obj.alpha0[index2] + (temp - obj.pi[index2])/(obj.pi[(index2 +
     1)] - obj.pi[index2]) * (obj.alpha0[(index2 +
     1)] - obj.alpha0[index2])
     new.y1 = K %*% (temp.alpha * obj$y) + temp.alpha0
     }
     }
     f.hat = cbind(f.hat, new.y1)
     new.y1 = as.numeric(new.y1 > 0) * 2 - 1
     pred.y = cbind(pred.y, new.y1)
     alpha = cbind(alpha, temp.alpha)
     alpha0 = c(alpha0, temp.alpha0)
     }
     colnames(f.hat) = NULL
     colnames(alpha) = NULL
     colnames(pred.y) = NULL
     z = list(pi = obj.pi, fitted.alpha0 = alpha0, fitted.alpha = alpha,
     fitted.f = f.hat, predicted.y = pred.y)
     return(z)
    }
    <bytecode: 0x30fd1a8>
    <environment: namespace:probsvm>
     --- function search by body ---
    Function predict_pipathresults in namespace probsvm has this body.
     ----------- END OF FAILURE REPORT --------------
    Error in if (class(new.x) != "matrix" & class(new.x) != "data.frame") { :
     the condition has length > 1
    Calls: probsvm -> loglik.svm -> prob.svm -> predict_pipathresults
    Execution halted
Flavor: r-devel-linux-x86_64-fedora-gcc