CRAN Package Check Results for Package msgl

Last updated on 2016-07-29 07:47:14.

Flavor Version Tinstall Tcheck Ttotal Status Flags
r-devel-linux-x86_64-debian-clang 2.2.0 24.24 68.39 92.63 OK
r-devel-linux-x86_64-debian-gcc 2.2.0 24.39 73.63 98.02 OK
r-devel-linux-x86_64-fedora-clang 2.2.0 188.53 OK
r-devel-linux-x86_64-fedora-gcc 2.2.0 171.96 OK
r-devel-osx-x86_64-clang 2.2.0 137.42 OK
r-devel-windows-ix86+x86_64 2.2.0 75.00 175.00 250.00 OK
r-patched-linux-x86_64 2.2.0 21.59 74.45 96.04 OK
r-patched-solaris-sparc 2.2.0 1065.90 OK
r-patched-solaris-x86 2.2.0 126.50 ERROR
r-release-linux-x86_64 2.2.0 19.83 75.83 95.65 OK
r-release-osx-x86_64-mavericks 2.2.0 OK
r-release-windows-ix86+x86_64 2.2.0 73.00 217.00 290.00 OK
r-oldrel-windows-ix86+x86_64 2.2.0 64.00 206.00 270.00 ERROR

Check Details

Version: 2.2.0
Check: examples
Result: ERROR
    Running examples in ‘msgl-Ex.R’ failed
    The error most likely occurred in:
    
    > ### Name: Err.msgl
    > ### Title: Compute error rates
    > ### Aliases: Err.msgl
    >
    > ### ** Examples
    >
    > data(SimData)
    > x.all <- sim.data$x
    > x.1 <- sim.data$x[1:50,]
    > x.2 <- sim.data$x[51:100,]
    > classes.all <- sim.data$classes
    > classes.1 <- sim.data$classes[1:50]
    > classes.2 <- sim.data$classes[51:100]
    >
    > #### Fit models using x.1
    > lambda <- msgl.lambda.seq(x.1, classes.1, alpha = .5, d = 25, lambda.min = 0.075)
    > fit <- msgl(x.1, classes.1, alpha = .5, lambda = lambda)
    
    Running msgl (dense design matrix)
    
     Samples: Features: Classes: Groups: Parameters:
     50 401 10 401 4.01k
    
    0 %
    |----|----|----|----|----|----|----|----|----|----|
    **************************************************|
    |
    >
    > #### Training errors:
    >
    > # Misclassification rate
    > Err(fit, x.1)
     [1] 0.90 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.68 0.68 0.66 0.60 0.58 0.56
    [16] 0.50 0.48 0.48 0.48 0.50 0.48 0.42 0.40 0.38 0.32
    >
    > # Misclassification count
    > Err(fit, x.1, type = "count")
     [1] 45 40 40 40 40 40 40 40 40 34 34 33 30 29 28 25 24 24 24 25 24 21 20 19 16
    >
    > # Negative log likelihood error
    > Err(fit, x.1, type="loglike")
     [1] 2.302585 2.297260 2.291854 2.286652 2.281604 2.276683 2.271869 2.267175
     [9] 2.262588 2.253558 2.242408 2.228843 2.213549 2.196275 2.178339 2.160463
    [17] 2.143003 2.125211 2.105954 2.086999 2.068345 2.049030 2.027378 2.006098
    [25] 1.981025
    >
    > # Misclassification rate of x.2
    > Err(fit, x.2, classes.2)
     [1] 0.90 0.88 0.88 0.88 0.88 0.88 0.90 0.90 0.90 0.90 0.88 0.86 0.80 0.78 0.74
    [16] 0.74 0.74 0.68 0.68 0.64 0.64 0.62 0.60 0.58 0.58
    >
    > #### Do cross validation
    > fit.cv <- msgl.cv(x.all, classes.all, alpha = .5, lambda = lambda)
    Running msgl 10 fold cross validation (dense design matrix)
    
     Samples: Features: Classes: Groups: Parameters:
     100 401 10 401 4.01k
    
    0 %
    |----|----|----|----|----|----|----|----|----|----|
    Error: C stack usage 191925568 is too close to the limit
    Error in sgl_subsampling(module_name, PACKAGE, data, parameterGrouping, :
     Aborted by user
    Calls: msgl.cv -> sgl_cv -> sgl_subsampling -> .Call
    Execution halted
Flavor: r-patched-solaris-x86

Version: 2.2.0
Check: tests
Result: ERROR
    Running the tests in ‘tests/msgl_cv_test_1.R’ failed.
    Last 13 lines of output:
     > fit.cv <- msgl.cv(x, classes, alpha = .5, lambda = lambda, standardize = TRUE, max.threads = threads)
     Running msgl 10 fold cross validation (dense design matrix)
    
     Samples: Features: Classes: Groups: Parameters:
     100 401 10 401 4.01k
    
     0 %
     |----|----|----|----|----|----|----|----|----|----|
     Error: C stack usage 191728448 is too close to the limit
     Error in sgl_subsampling(module_name, PACKAGE, data, parameterGrouping, :
     Aborted by user
     Calls: msgl.cv -> sgl_cv -> sgl_subsampling -> .Call
     Execution halted
Flavor: r-patched-solaris-x86

Version: 2.2.0
Check: running tests for arch ‘i386’
Result: ERROR
    Running the tests in 'tests/msgl_grouping_test_1.R' failed.
    Last 13 lines of output:
     > lambda1 <- msgl.lambda.seq(x, classes, grouping = grouping, alpha = 0, d = 100L, lambda.min = 0.01, sparse.data = TRUE, standardize = FALSE)
     > if(max(abs(lambda-lambda1)) > 1e-5) stop()
     >
     > lambda <- msgl.lambda.seq(x, classes, grouping = grouping, alpha = .5, d = 100L, lambda.min = 0.01, standardize = FALSE)
     Error in sgl_lambda_sequence("msgl_dense", "msgl", data, covariateGrouping, :
     The algorithm has encountered a numerical problem
     Try/check the following:
     (1) Check the lambda sequence and try with a longer sequence (the length d should be larger than 100)
     (2) try with a larger lambda.min
     (3) Ensure minimum one sample from each group/class present in all subsample (or cross validation) sets used
     (Assert failed in d:/RCompile/CRANpkg/lib/3.2/sglOptim/include/sgl/sgl_problem.h at line 379 )
     Calls: msgl.lambda.seq -> sgl_lambda_sequence -> .Call
     Execution halted
Flavor: r-oldrel-windows-ix86+x86_64