sumbose/irf

iterative Random Forests (iRF): iteratively grows weighted random forests, finds interaction among features

irf's Issues

[JOSS review] Add code of conduct and contributor guidelines

JOSS requires contributor guidelines and code of conduct to be available. You can use usethis::use_code_of_conduct and usethis::use_tidy_contributing to add them (and please link them in the README).

[JOSS review] Add DOI to references

Please add the corresponding DOI to the references (if available) in paper.bib. Format example here.

openjournals/joss-reviews#1077

FR: Provide wrappers to use with `caret` package

It would be helpful to have wrappers to allow iRF to be used with caret, which provides a consistent interface to a number of ML methods.

• GitHub: https://github.com/topepo/caret
• Documentation: http://topepo.github.io/caret/index.html

[JOSS review] Use detectCores() instead of specific number in example

Currently, the code in the vignette uses n.cores <- 4. I'd recommend replacing it with the more flexible n.cores <- detectCores() so that users with different numbers of cpu cores can achieve maximal parallelization on their machines.

openjournals/joss-reviews#1077

[JOSS review] Installation guideline improvements

Some items in the readme on installation could be improved:

1. It looks like the package is already available from CRAN (awesome!), so it is usually good practice to include the installation guidelines for the CRAN version in addition to the GitHub version. A simple example: https://github.com/road2stat/liftr#installation

2. I would also suggest removing the "Alternatively, ... R CMD INSTALL" method because it is more manual and not too different from using devtools::install_github, at least for regular users who don't want to modify the source code before using the software.

3. gfortran. I'd probably mention that to compile iRF from source, it needs gfortran, and point to the official CRAN page https://cran.r-project.org/bin/macosx/tools/ for users who want to install it under MacOS. I would also remove the command line since it doesn't handle the complexity of system versions and gfortran versions.

openjournals/joss-reviews#1077

[JOSS review] Add some unit tests

JOSS requires some automated tests or manual test instructions available in the software. Please consider following the guidelines here to create some basic unit tests for certain functions in the package.

[JOSS review] Fix GitHub version installation issue

Currently, when running devtools::install_github("sumbose/iRF"), I'm seeing

** testing if installed package can be loaded
Error: package or namespace load failed for ‘iRF’ in namespaceExport(ns, exports):
 undefined exports: interactPredict
Error: loading failed
Execution halted

Should be a quick fix.

openjournals/joss-reviews#1077

Two questions about parameters

What are the criteria for n_trees of RIT and n.bootstrap inside the irf function?

[JOSS review] Fix affiliation format in paper.md

It seems the format of two affiliations in paper.mdis not valid YAML... should be in a single line. That's probably why whedon had troubles generating pdf.

To fix this, please merge line 25 and 26; also line 28 and 29, so the paper can be compiled properly.

openjournals/joss-reviews#1077

[JOSS review] Fix example code in vignette

Currently, when using the CRAN version and running this code block in the vignette:

> rforest$node.feature[1:10, c(1:5, p + 1:5)]

it shows:

Error in intI(j, n = d[2], dn[[2]], give.dn = FALSE) : 
  index larger than maximal 250

and almost everything after this also breaks.

openjournals/joss-reviews#1077

Error in unserialize(socklist[[n]]) : error reading from connection

Hi,
Thank you for your package.

When I applyed function "iRF", an error has been reported as "Error in unserialize(socklist[[n]]) : error reading from connection".

The codes I ran are as following:

fit <- iRF(x=as.matrix(data[train.id, exposure]),
y=data[train.id,outcome],
xtest=as.matrix(data[test.id, exposure]),
ytest=data[test.id,outcome],
n.iter=10,
n.core=3,
select.iter = T,
n.bootstrap=200
)

I ran the code on windows on my own laptop. I was wondering what are the possible reasons for this？

Looking forward to your answer.

Cannot install iRF

Hey Sumanta!

I wanted to use iRF to identify interactions between some variables for one of my projects, but I can't seem to install the package. Not sure if it's supposed to be usable yet, but I just thought I'd let you know.

> devtools::install_github("sumbose/iRF")
Downloading GitHub repo sumbose/iRF@master
from URL https://api.github.com/repos/sumbose/iRF/zipball/master
Installing iRF
'/Library/Frameworks/R.framework/Resources/bin/R' --no-site-file --no-environ --no-save --no-restore --quiet CMD INSTALL  \
  '/private/var/folders/yv/wgyt24cx34j3fyxz_y4lzx380000gn/T/RtmpDc0iIZ/devtoolsc5c731328a1e/sumbose-iRF-062fb3f'  \
  --library='/Library/Frameworks/R.framework/Versions/3.3/Resources/library' --install-tests 

* installing *source* package ‘iRF’ ...
** libs
clang -I/Library/Frameworks/R.framework/Resources/include -DNDEBUG  -I/usr/local/include -I/usr/local/include/freetype2 -I/opt/X11/include    -fPIC  -Wall -mtune=core2 -g -O2  -c classTree.c -o classTree.o
clang -I/Library/Frameworks/R.framework/Resources/include -DNDEBUG  -I/usr/local/include -I/usr/local/include/freetype2 -I/opt/X11/include    -fPIC  -Wall -mtune=core2 -g -O2  -c regTree.c -o regTree.o
regTree.c:184:9: warning: unused variable 'jj' [-Wunused-variable]
    int jj, jold;
        ^
regTree.c:391:14: warning: unused variable 'dum' [-Wunused-variable]
        int i, tmp, dum;
                    ^
2 warnings generated.
clang -I/Library/Frameworks/R.framework/Resources/include -DNDEBUG  -I/usr/local/include -I/usr/local/include/freetype2 -I/opt/X11/include    -fPIC  -Wall -mtune=core2 -g -O2  -c regrf.c -o regrf.o
regrf.c:67:9: warning: unused variable 'ii' [-Wunused-variable]
    int ii, jj, tmpcheck, subsetCard;
        ^
regrf.c:67:13: warning: unused variable 'jj' [-Wunused-variable]
    int ii, jj, tmpcheck, subsetCard;
            ^
regrf.c:67:17: warning: unused variable 'tmpcheck' [-Wunused-variable]
    int ii, jj, tmpcheck, subsetCard;
                ^
3 warnings generated.
clang -I/Library/Frameworks/R.framework/Resources/include -DNDEBUG  -I/usr/local/include -I/usr/local/include/freetype2 -I/opt/X11/include    -fPIC  -Wall -mtune=core2 -g -O2  -c rf.c -o rf.o
rf.c:471:1: warning: expression result unused [-Wunused-value]
+                                                               nrightimp[cl[n] - 1]++;
^                                                               ~~~~~~~~~~~~~~~~~~~~~~
rf.c:100:13: warning: unused variable 'jj' [-Wunused-variable]
    int ii, jj, tmpcheck, subsetCard;
            ^
rf.c:100:9: warning: unused variable 'ii' [-Wunused-variable]
    int ii, jj, tmpcheck, subsetCard;
        ^
3 warnings generated.
gfortran-4.8   -fPIC  -g -O2  -c rfsub.f -o rfsub.o
make: gfortran-4.8: No such file or directory
make: *** [rfsub.o] Error 1
ERROR: compilation failed for package ‘iRF’
* removing ‘/Library/Frameworks/R.framework/Versions/3.3/Resources/library/iRF’
Error: Command failed (1)

My sessionInfo() is as follows:

> sessionInfo()
R version 3.3.2 (2016-10-31)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: macOS Sierra 10.12.3

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
[1] httr_1.2.1      R6_2.2.0        tools_3.3.2     withr_1.0.2     curl_2.3        memoise_1.0.0   git2r_0.18.0    digest_0.6.12   devtools_1.12.0

[JOSS review] Make vignette available online

Some documentation improvements:

1. It is often very useful to make vignette directly readable online. The fastest way I can think of is using pkgdown and host the website with GitHub Pages.

2. When it's ready, please add the website or vignette link (if you use GitHub Pages, it would be https://sumbose.github.io/iRF/ ) to the description section of the repo (click the "Edit" button and fill in the website address then click "Save").

3. Add the link of the online vignette to the README.md so users know where to look to get started.

openjournals/joss-reviews#1077

Cannot install iRF from github on MacOS

Hello, Sumanta,

I want to install iRF on MacOS, and I already install gfortran to compile, but I still cannot successfully use this package, and it will show the following error after install:

> devtools::install_github("sumbose/iRF")
Downloading GitHub repo sumbose/iRF@master
✔  checking for file ‘/private/var/folders/g2/fknsdml54vsfn7k25q_j6p3h0000gn/T/RtmpQMSNbU/remotes1084d68ac7f1a/sumbose-iRF-fafed6a/DESCRIPTION’ ...
─  preparing ‘iRF’:
✔  checking DESCRIPTION meta-information ...
─  cleaning src
─  checking for LF line-endings in source and make files and shell scripts
─  checking for empty or unneeded directories
─  looking to see if a ‘data/datalist’ file should be added
─  building ‘iRF_3.0.0.tar.gz’
   
* installing *source* package ‘iRF’ ...
** libs
clang++ -std=gnu++11 -I"/Library/Frameworks/R.framework/Resources/include" -DNDEBUG  -I"/Library/Frameworks/R.framework/Versions/3.5/Resources/library/Rcpp/include" -I/usr/local/include  -fopenmp -fPIC  -Wall -g -O2 -c ExportedFunctionsRIT.cpp -o ExportedFunctionsRIT.o
clang: error: unsupported option '-fopenmp'
make: *** [ExportedFunctionsRIT.o] Error 1
ERROR: compilation failed for package ‘iRF’
* removing ‘/Library/Frameworks/R.framework/Versions/3.5/Resources/library/iRF’
* restoring previous ‘/Library/Frameworks/R.framework/Versions/3.5/Resources/library/iRF’
Error: Failed to install 'iRF' from GitHub:
  (converted from warning) installation of package ‘/var/folders/g2/fknsdml54vsfn7k25q_j6p3h0000gn/T//RtmpQMSNbU/file1084da4ac855/iRF_3.0.0.tar.gz’ had non-zero exit status

My sessionInfo() is as follow:

> sessionInfo()
R version 3.5.2 (2018-12-20)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Mojave 10.14.6

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] iRF_2.0.0

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.2         pillar_1.4.2       compiler_3.5.2     RColorBrewer_1.1-2 remotes_2.1.0     
 [6] prettyunits_1.0.2  iterators_1.0.12   tools_3.5.2        testthat_2.2.1     digest_0.6.21     
[11] pkgbuild_1.0.3     pkgload_1.0.2      memoise_1.1.0      tibble_2.1.3       lattice_0.20-38   
[16] pkgconfig_2.0.3    rlang_0.4.0        Matrix_1.2-17      foreach_1.4.7      cli_1.1.0         
[21] rstudioapi_0.10    curl_3.3           parallel_3.5.2     withr_2.1.2        dplyr_0.8.3       
[26] fs_1.3.1           desc_1.2.0         AUC_0.3.0          devtools_2.2.1     rprojroot_1.3-2   
[31] grid_3.5.2         tidyselect_0.2.5   glue_1.3.1         data.table_1.12.2  R6_2.4.0          
[36] processx_3.3.1     sessioninfo_1.1.1  purrr_0.3.2        callr_3.2.0        magrittr_1.5      
[41] usethis_1.5.0      backports_1.1.4    codetools_0.2-16   ps_1.3.0           ellipsis_0.3.0    
[46] assertthat_0.2.1   doParallel_1.0.15  crayon_1.3.4  

Failed to install from Github

Thanks for the wonderful package.

Got the following error when trying to install from Github:

#error This file requires compiler and library support for the ISO C++ 2011 standard. This support must be enabled with the -std=c++11 or -std=gnu++11 compiler options.

Solved the issue by

1. create a file named "CMakeLists.txt" in the src directory.
2. Put this line:

set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")

on the second line of the newly created CMakeLists.txt file.

Warning: "condition has length > 1 and only the first element will be used"

Using iRF version 3.0.0 installed yesterday from GitHub, I get this warning message:

Warning message:
In if (!class(x) %in% c("data.frame", "matrix")) { :
  the condition has length > 1 and only the first element will be used

when x is of class c("matrix", "array"). This is originating from

iRF/R/iRF.R

Lines 93 to 97 in 7476536

	if (!class(x) %in% c('data.frame', 'matrix')) {
	sp.mat <- attr(class(x), 'package') == 'Matrix'
	if (is.null(sp.mat) || !sp.mat)
	stop('x must be matrix or data frame')
	}

This can be resolved by replacing class(x) %in% c("data.frame", "matrix") with is(x, c("data.frame", "matrix") here, as well as elsewhere in the code.

Example:

set.seed(42)
x <- structure(c(0.51542, 0.61489, 0.63492, 0.53612, 0.52895, 0.64572,  0.54093, 0.48915, 0.56444, 0.62533, 0.48457, 0.52632, 0.52961,  0.65015, 0.54746, 0.53074, 0.54062, 0.51413, 0.58454, 0.57131,  0.26293, 0.29376, 0.28257, 0.27655, 0.27326, 0.23358, 0.28784,  0.31006, 0.28066, 0.25501, 0.28015, 0.2786, 0.24955, 0.27152,  0.2051, 0.20854, 0.28254, 0.30097, 0.27195, 0.24717), .Dim = c(10L,  4L), .Dimnames = list(NULL, c("CGI_69.PDR.Set1", "CGI_223.PDR.Set1",  "CGI_299.PDR.Set1", "CGI_550.PDR.Set1")))
 y <- sample(factor(c("0","1")), size=10, replace=TRUE)

r <- iRF(x=x, y=y)
## [1] "iteration =  1"
## [1] "iteration =  2"
## [1] "iteration =  3"
## [1] "iteration =  4"
## [1] "iteration =  5"
## Warning message:
## In if (!class(x) %in% c("data.frame", "matrix")) { :
##   the condition has length > 1 and only the first element will be used## Warning message:

object 'X' not found

Hi,

When running the iRF function, I found that the training variable x is referred to in the body of the function as X, this is line 24 and 25 in the iRF.R code. Is this a typo?

P