[bug#48388] [PATCH 0/3] Add xgboost.

[bug#48388] [PATCH 0/3] Add xgboost.

[Vinicius Monego]

This patchset adds the binaries, C++ library and Python frontend for the XGBoost gradient boosting framework.

I noticed r-xgboost in the packages, which compiles a bundled copy of xgboost. Ideally it would inherit from xgboost and build from there, but I do not work with R so I can't help with this.

I also had to do some dirty hack in the python patch for reading version and using libxgboost from the xgboost package. Lmk is there is a better way to handle this.

Vinicius Monego (3):
  gnu: Add dmlc-core.
  gnu: Add xgboost.
  gnu: Add python-xgboost.

 gnu/local.mk                                  |   2 +
 gnu/packages/machine-learning.scm             | 125 ++++++++++++++++++
 ...xgboost-python-use-system-libxgboost.patch |  69 ++++++++++
 .../xgboost-use-system-dmlc-core.patch        |  35 +++++
 4 files changed, 231 insertions(+)
 create mode 100644 gnu/packages/patches/xgboost-python-use-system-libxgboost.patch
 create mode 100644 gnu/packages/patches/xgboost-use-system-dmlc-core.patch

-- 
2.31.1

[bug#48388] [PATCH 1/3] gnu: Add dmlc-core.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (dmlc-core): New variable.
---
 gnu/packages/machine-learning.scm | 32 +++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 837ebbba67..9d6c5505f7 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1927,6 +1927,38 @@ together building blocks and a subclassing API with an imperative style for
 advanced research.")
     (license license:asl2.0)))
 
+(define-public dmlc-core
+  ;; Version 0.4 is too old for xgboost.
+  (let ((version "0.5")
+        (commit "f00e3ec7abc9f293a1b7061157b0a4e22a735cf5")
+        (revision "0"))
+    (package
+      (name "dmlc-core")
+      (version (git-version version revision commit))
+      (source
+       (origin
+         (method git-fetch)
+         (uri (git-reference
+               (url "https://github.com/dmlc/dmlc-core")
+               (commit commit)))
+         (file-name (git-file-name name version))
+         (sha256
+          (base32 "1x4ad1jhn84fywlk031fmv1kxyiscclmrqn9hhj8gz0mh7z9vcrh"))))
+      (build-system cmake-build-system)
+      (arguments
+       `(#:configure-flags
+         (list "-DGOOGLE_TEST=ON")))
+      (native-inputs
+       `(("googletest" ,googletest)
+         ("python" ,python-wrapper)))
+      (home-page "https://github.com/dmlc/dmlc-core")
+      (synopsis "Common bricks library for machine learning")
+      (description
+       "DMLC-Core is the backbone library to support all DMLC projects,
+offers the bricks to build efficient and scalable distributed machine
+learning libraries.")
+      (license license:asl2.0))))
+
 (define-public python-iml
   (package
     (name "python-iml")
-- 
2.31.1

[bug#48388] [PATCH 2/3] gnu: Add xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (xgboost): New variable.
---
 gnu/local.mk                                  |  2 +
 gnu/packages/machine-learning.scm             | 33 +++++++++
 ...xgboost-python-use-system-libxgboost.patch | 69 +++++++++++++++++++
 .../xgboost-use-system-dmlc-core.patch        | 35 ++++++++++
 4 files changed, 139 insertions(+)
 create mode 100644 gnu/packages/patches/xgboost-python-use-system-libxgboost.patch
 create mode 100644 gnu/packages/patches/xgboost-use-system-dmlc-core.patch

diff --git a/gnu/local.mk b/gnu/local.mk
index 37166bb2fc..e1d382c543 100644
--- a/gnu/local.mk
+++ b/gnu/local.mk
@@ -1823,6 +1823,8 @@ dist_patch_DATA =						\
   %D%/packages/patches/xf86-video-voodoo-pcitag.patch		\
   %D%/packages/patches/xfce4-panel-plugins.patch		\
   %D%/packages/patches/xfce4-settings-defaults.patch		\
+  %D%/packages/patches/xgboost-python-use-system-libxgboost.patch \
+  %D%/packages/patches/xgboost-use-system-dmlc-core.patch       \
   %D%/packages/patches/xmonad-dynamic-linking.patch		\
   %D%/packages/patches/xplanet-1.3.1-cxx11-eof.patch		\
   %D%/packages/patches/xplanet-1.3.1-libdisplay_DisplayOutput.cpp.patch	\
diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 9d6c5505f7..fd10de427f 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1959,6 +1959,39 @@ offers the bricks to build efficient and scalable distributed machine
 learning libraries.")
       (license license:asl2.0))))
 
+(define-public xgboost
+  (package
+    (name "xgboost")
+    (version "1.4.1")
+    (source
+     (origin
+       (method git-fetch)
+       (uri (git-reference
+             (url "https://github.com/dmlc/xgboost")
+             (commit (string-append "v" version))))
+       (file-name (git-file-name name version))
+       (patches (search-patches "xgboost-python-use-system-libxgboost.patch"
+                                "xgboost-use-system-dmlc-core.patch"))
+       (sha256
+        (base32 "0ixacmpcj844c09xigi1vd4rrckld84k32iarvhnvvm6yjih756y"))))
+    (build-system cmake-build-system)
+    (arguments
+     `(#:configure-flags (list "-DGOOGLE_TEST=ON")))
+    (native-inputs
+     `(("googletest" ,googletest)
+       ("python" ,python-wrapper)))
+    (inputs
+     `(("dmlc-core" ,dmlc-core)))
+    (home-page "https://xgboost.ai/")
+    (synopsis "Gradient boosting (GBDT, GBRT or GBM) library in C++")
+    (description
+     "XGBoost is an optimized distributed gradient boosting library designed
+to be highly efficient, flexible and portable.  It implements machine learning
+algorithms under the Gradient Boosting framework.  XGBoost provides a parallel
+tree boosting (also known as GBDT, GBM) that solve many data science problems
+in a fast and accurate way.")
+    (license license:asl2.0)))
+
 (define-public python-iml
   (package
     (name "python-iml")
diff --git a/gnu/packages/patches/xgboost-python-use-system-libxgboost.patch b/gnu/packages/patches/xgboost-python-use-system-libxgboost.patch
new file mode 100644
index 0000000000..992e6bb3fd
--- /dev/null
+++ b/gnu/packages/patches/xgboost-python-use-system-libxgboost.patch
@@ -0,0 +1,69 @@
+From 768192b9c6700f0ae6b750c56b114d7022692a0b Mon Sep 17 00:00:00 2001
+From: Vinicius Monego <monego@posteo.net>
+Date: Tue, 11 May 2021 19:30:55 -0300
+Subject: [PATCH] Python wrapper: use system libxgboost.
+
+---
+This patch was extended from the following Debian patch: https://sources.debian.org/src/xgboost/1.2.1-1/debian/patches/setup.py.patch/
+
+ python-package/setup.py            | 4 ++--
+ python-package/xgboost/__init__.py | 5 ++---
+ python-package/xgboost/libpath.py  | 5 +++++
+ 3 files changed, 9 insertions(+), 5 deletions(-)
+
+diff --git a/python-package/setup.py b/python-package/setup.py
+index a2feb766..b8b78429 100644
+--- a/python-package/setup.py
++++ b/python-package/setup.py
+@@ -312,11 +312,11 @@ if __name__ == '__main__':
+               'numpy',
+               'scipy',
+           ],
+-          ext_modules=[CMakeExtension('libxgboost')],
++          # ext_modules=[CMakeExtension('libxgboost')],
+           cmdclass={
+               'build_ext': BuildExt,
+               'sdist': Sdist,
+-              'install_lib': InstallLib,
++              # 'install_lib': InstallLib,
+               'install': Install
+           },
+           extras_require={
+diff --git a/python-package/xgboost/__init__.py b/python-package/xgboost/__init__.py
+index e0ff434d..658e84e6 100644
+--- a/python-package/xgboost/__init__.py
++++ b/python-package/xgboost/__init__.py
+@@ -21,9 +21,8 @@ try:
+ except ImportError:
+     pass
+ 
+-VERSION_FILE = os.path.join(os.path.dirname(__file__), 'VERSION')
+-with open(VERSION_FILE) as f:
+-    __version__ = f.read().strip()
++# This variable will be set during substitution.
++__version__ = ''
+ 
+ __all__ = ['DMatrix', 'DeviceQuantileDMatrix', 'Booster',
+            'train', 'cv',
+diff --git a/python-package/xgboost/libpath.py b/python-package/xgboost/libpath.py
+index f7a7d9cd..88d9d986 100644
+--- a/python-package/xgboost/libpath.py
++++ b/python-package/xgboost/libpath.py
+@@ -19,9 +19,14 @@ def find_lib_path() -> List[str]:
+     lib_path
+        List of all found library path to xgboost
+     """
++
++    # This variable will be set during substitution.
++    XGBOOST_PATH = ''
++
+     curr_path = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
+     dll_path = [
+         # normal, after installation `lib` is copied into Python package tree.
++        os.path.join(XGBOOST_PATH, 'lib'),
+         os.path.join(curr_path, 'lib'),
+         # editable installation, no copying is performed.
+         os.path.join(curr_path, os.path.pardir, os.path.pardir, 'lib'),
+-- 
+2.31.1
+
diff --git a/gnu/packages/patches/xgboost-use-system-dmlc-core.patch b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
new file mode 100644
index 0000000000..6209ae1f10
--- /dev/null
+++ b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
@@ -0,0 +1,35 @@
+From 34167cc105e47589b86b6f62da76e0ab744002dd Mon Sep 17 00:00:00 2001
+From: Vinicius Monego <monego@posteo.net>
+Date: Tue, 11 May 2021 19:35:30 -0300
+Subject: [PATCH] Use dmlc-core from Guix.
+
+---
+This patch is a subset from the following Debian patch: https://sources.debian.org/src/xgboost/1.2.1-1/debian/patches/cmake.patch/
+
+Rabit is now developed as part of xgboost, so we don't build it separately.
+
+ CMakeLists.txt | 8 +++-----
+ 1 file changed, 3 insertions(+), 5 deletions(-)
+
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 6cc19fbd..3b42bc17 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -150,11 +150,9 @@ endif (USE_OPENMP)
+ 
+ # dmlc-core
+ msvc_use_static_runtime()
+-add_subdirectory(${xgboost_SOURCE_DIR}/dmlc-core)
+-set_target_properties(dmlc PROPERTIES
+-  CXX_STANDARD 14
+-  CXX_STANDARD_REQUIRED ON
+-  POSITION_INDEPENDENT_CODE ON)
++add_library(dmlc SHARED IMPORTED)
++find_library(DMLC_LIBRARY dmlc)
++set_property(TARGET dmlc PROPERTY IMPORTED_LOCATION "${DMLC_LIBRARY}")
+ if (MSVC)
+   target_compile_options(dmlc PRIVATE
+                          -D_CRT_SECURE_NO_WARNINGS -D_CRT_SECURE_NO_DEPRECATE)
+-- 
+2.31.1
+
-- 
2.31.1

[bug#48388] [PATCH 3/3] gnu: Add python-xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (python-xgboost): New variable.
---
 gnu/packages/machine-learning.scm | 60 +++++++++++++++++++++++++++++++
 1 file changed, 60 insertions(+)

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index fd10de427f..f7bb01f274 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1992,6 +1992,66 @@ tree boosting (also known as GBDT, GBM) that solve many data science problems
 in a fast and accurate way.")
     (license license:asl2.0)))
 
+(define-public python-xgboost
+  (package
+    (inherit xgboost)
+    (name "python-xgboost")
+    (build-system python-build-system)
+    (arguments
+     `(#:phases
+       (modify-phases %standard-phases
+         (add-after 'unpack 'preparations
+           (lambda* (#:key inputs #:allow-other-keys)
+             ;; Move python-package content to parent directory to silence
+             ;; some warnings about files not being found if we chdir.
+             (rename-file "python-package/xgboost" "xgboost")
+             (rename-file "python-package/README.rst" "README.rst")
+             (rename-file "python-package/setup.cfg" "setup.cfg")
+             (rename-file "python-package/setup.py" "setup.py")
+
+             (with-directory-excursion "xgboost"
+               ;; Version is read from a VERSION file, but that file is not
+               ;; part of the installation.
+               (substitute* "__init__.py"
+                 (("__version__ = ''")
+                  (string-append "__version__ = " "'"
+                                 ,(package-version xgboost) "'")))
+               ;; The Python wrapper expects libxgboost to be found in the
+               ;; same source from where the wrapper builds.  This workaround
+               ;; allows libxgboost to be used from the xgboost package.
+               (substitute* "libpath.py"
+                 (("XGBOOST_PATH = ''")
+                  (string-append "XGBOOST_PATH = "
+                                 "'" (assoc-ref inputs "xgboost") "'"))))
+             #t))
+         (replace 'check
+           ;; Python-specific tests are located in tests/python.
+           (lambda* (#:key inputs outputs #:allow-other-keys)
+             (add-installed-pythonpath inputs outputs)
+             (invoke "pytest" "tests/python"
+                     ;; FIXME: CLI tests fail with PermissionError.
+                     "--ignore" "tests/python/test_cli.py" "-k"
+                     (string-append
+                      "not test_cli_regression_demo"
+                      ;; The tests below open a network connection.
+                      " and not test_model_compatibility"
+                      " and not test_get_group"
+                      " and not test_cv_no_shuffle"
+                      " and not test_cv"
+                      " and not test_training"
+                      ;; "'['./runexp.sh']' returned non-zero exit status 1."
+                      " and not test_cli_binary_classification")))))))
+    (native-inputs
+     `(("python-pandas" ,python-pandas)
+       ("python-pytest" ,python-pytest)
+       ("python-scikit-learn" ,python-scikit-learn)))
+    (inputs
+     `(("xgboost" ,xgboost)))
+    (propagated-inputs
+     `(("python-numpy" ,python-numpy)
+       ("python-scipy" ,python-scipy)))
+    (synopsis "Python interface for the XGBoost library")))
+
 (define-public python-iml
   (package
     (name "python-iml")
-- 
2.31.1

[bug#48388] [PATCH v2 2/3] gnu: Add xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (xgboost): New variable.
* gnu/packages/patches/xgboost-python-use-system-libxgboost.patch,
gnu/packages/patches/xgboost-use-system-dmlc-core.patch: New files.
* gnu/local.mk (dist_PATCH_DATA): Add them.
---
Changed commit message to add patch files.

 gnu/local.mk                                  |  2 +
 gnu/packages/machine-learning.scm             | 33 +++++++++
 ...xgboost-python-use-system-libxgboost.patch | 69 +++++++++++++++++++
 .../xgboost-use-system-dmlc-core.patch        | 35 ++++++++++
 4 files changed, 139 insertions(+)
 create mode 100644 gnu/packages/patches/xgboost-python-use-system-libxgboost.patch
 create mode 100644 gnu/packages/patches/xgboost-use-system-dmlc-core.patch

diff --git a/gnu/local.mk b/gnu/local.mk
index 37166bb2fc..e1d382c543 100644
--- a/gnu/local.mk
+++ b/gnu/local.mk
@@ -1823,6 +1823,8 @@ dist_patch_DATA =						\
   %D%/packages/patches/xf86-video-voodoo-pcitag.patch		\
   %D%/packages/patches/xfce4-panel-plugins.patch		\
   %D%/packages/patches/xfce4-settings-defaults.patch		\
+  %D%/packages/patches/xgboost-python-use-system-libxgboost.patch \
+  %D%/packages/patches/xgboost-use-system-dmlc-core.patch       \
   %D%/packages/patches/xmonad-dynamic-linking.patch		\
   %D%/packages/patches/xplanet-1.3.1-cxx11-eof.patch		\
   %D%/packages/patches/xplanet-1.3.1-libdisplay_DisplayOutput.cpp.patch	\
diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 9d6c5505f7..d81370797f 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1959,6 +1959,39 @@ offers the bricks to build efficient and scalable distributed machine
 learning libraries.")
       (license license:asl2.0))))
 
+(define-public xgboost
+  (package
+    (name "xgboost")
+    (version "1.4.1")
+    (source
+     (origin
+       (method git-fetch)
+       (uri (git-reference
+             (url "https://github.com/dmlc/xgboost")
+             (commit (string-append "v" version))))
+       (file-name (git-file-name name version))
+       (patches (search-patches "xgboost-python-use-system-libxgboost.patch"
+                                "xgboost-use-system-dmlc-core.patch"))
+       (sha256
+        (base32 "0ixacmpcj844c09xigi1vd4rrckld84k32iarvhnvvm6yjih756y"))))
+    (build-system cmake-build-system)
+    (arguments
+     `(#:configure-flags (list "-DGOOGLE_TEST=ON")))
+    (native-inputs
+     `(("googletest" ,googletest)
+       ("python" ,python-wrapper)))
+    (inputs
+     `(("dmlc-core" ,dmlc-core)))
+    (home-page "https://xgboost.ai/")
+    (synopsis "Gradient boosting (GBDT, GBRT or GBM) library in C++")
+    (description
+     "XGBoost is an optimized distributed gradient boosting library designed
+to be highly efficient, flexible and portable.  It implements machine learning
+algorithms under the Gradient Boosting framework.  XGBoost provides a parallel
+tree boosting (also known as GBDT, GBM) that solve many data science problems
+in a fast and accurate way.")
+    (license license:asl2.0)))
+
 (define-public python-iml
   (package
     (name "python-iml")
diff --git a/gnu/packages/patches/xgboost-python-use-system-libxgboost.patch b/gnu/packages/patches/xgboost-python-use-system-libxgboost.patch
new file mode 100644
index 0000000000..992e6bb3fd
--- /dev/null
+++ b/gnu/packages/patches/xgboost-python-use-system-libxgboost.patch
@@ -0,0 +1,69 @@
+From 768192b9c6700f0ae6b750c56b114d7022692a0b Mon Sep 17 00:00:00 2001
+From: Vinicius Monego <monego@posteo.net>
+Date: Tue, 11 May 2021 19:30:55 -0300
+Subject: [PATCH] Python wrapper: use system libxgboost.
+
+---
+This patch was extended from the following Debian patch: https://sources.debian.org/src/xgboost/1.2.1-1/debian/patches/setup.py.patch/
+
+ python-package/setup.py            | 4 ++--
+ python-package/xgboost/__init__.py | 5 ++---
+ python-package/xgboost/libpath.py  | 5 +++++
+ 3 files changed, 9 insertions(+), 5 deletions(-)
+
+diff --git a/python-package/setup.py b/python-package/setup.py
+index a2feb766..b8b78429 100644
+--- a/python-package/setup.py
++++ b/python-package/setup.py
+@@ -312,11 +312,11 @@ if __name__ == '__main__':
+               'numpy',
+               'scipy',
+           ],
+-          ext_modules=[CMakeExtension('libxgboost')],
++          # ext_modules=[CMakeExtension('libxgboost')],
+           cmdclass={
+               'build_ext': BuildExt,
+               'sdist': Sdist,
+-              'install_lib': InstallLib,
++              # 'install_lib': InstallLib,
+               'install': Install
+           },
+           extras_require={
+diff --git a/python-package/xgboost/__init__.py b/python-package/xgboost/__init__.py
+index e0ff434d..658e84e6 100644
+--- a/python-package/xgboost/__init__.py
++++ b/python-package/xgboost/__init__.py
+@@ -21,9 +21,8 @@ try:
+ except ImportError:
+     pass
+ 
+-VERSION_FILE = os.path.join(os.path.dirname(__file__), 'VERSION')
+-with open(VERSION_FILE) as f:
+-    __version__ = f.read().strip()
++# This variable will be set during substitution.
++__version__ = ''
+ 
+ __all__ = ['DMatrix', 'DeviceQuantileDMatrix', 'Booster',
+            'train', 'cv',
+diff --git a/python-package/xgboost/libpath.py b/python-package/xgboost/libpath.py
+index f7a7d9cd..88d9d986 100644
+--- a/python-package/xgboost/libpath.py
++++ b/python-package/xgboost/libpath.py
+@@ -19,9 +19,14 @@ def find_lib_path() -> List[str]:
+     lib_path
+        List of all found library path to xgboost
+     """
++
++    # This variable will be set during substitution.
++    XGBOOST_PATH = ''
++
+     curr_path = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
+     dll_path = [
+         # normal, after installation `lib` is copied into Python package tree.
++        os.path.join(XGBOOST_PATH, 'lib'),
+         os.path.join(curr_path, 'lib'),
+         # editable installation, no copying is performed.
+         os.path.join(curr_path, os.path.pardir, os.path.pardir, 'lib'),
+-- 
+2.31.1
+
diff --git a/gnu/packages/patches/xgboost-use-system-dmlc-core.patch b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
new file mode 100644
index 0000000000..6209ae1f10
--- /dev/null
+++ b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
@@ -0,0 +1,35 @@
+From 34167cc105e47589b86b6f62da76e0ab744002dd Mon Sep 17 00:00:00 2001
+From: Vinicius Monego <monego@posteo.net>
+Date: Tue, 11 May 2021 19:35:30 -0300
+Subject: [PATCH] Use dmlc-core from Guix.
+
+---
+This patch is a subset from the following Debian patch: https://sources.debian.org/src/xgboost/1.2.1-1/debian/patches/cmake.patch/
+
+Rabit is now developed as part of xgboost, so we don't build it separately.
+
+ CMakeLists.txt | 8 +++-----
+ 1 file changed, 3 insertions(+), 5 deletions(-)
+
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 6cc19fbd..3b42bc17 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -150,11 +150,9 @@ endif (USE_OPENMP)
+ 
+ # dmlc-core
+ msvc_use_static_runtime()
+-add_subdirectory(${xgboost_SOURCE_DIR}/dmlc-core)
+-set_target_properties(dmlc PROPERTIES
+-  CXX_STANDARD 14
+-  CXX_STANDARD_REQUIRED ON
+-  POSITION_INDEPENDENT_CODE ON)
++add_library(dmlc SHARED IMPORTED)
++find_library(DMLC_LIBRARY dmlc)
++set_property(TARGET dmlc PROPERTY IMPORTED_LOCATION "${DMLC_LIBRARY}")
+ if (MSVC)
+   target_compile_options(dmlc PRIVATE
+                          -D_CRT_SECURE_NO_WARNINGS -D_CRT_SECURE_NO_DEPRECATE)
+-- 
+2.31.1
+
-- 
2.31.1

[bug#48388] [PATCH v2 3/3] gnu: Add python-xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (python-xgboost): New variable.
---
Added (when tests?) in check phase to allow --without-tests.

 gnu/packages/machine-learning.scm | 62 +++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index d81370797f..ead47a534a 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1992,6 +1992,68 @@ tree boosting (also known as GBDT, GBM) that solve many data science problems
 in a fast and accurate way.")
     (license license:asl2.0)))
 
+(define-public python-xgboost
+  (package
+    (inherit xgboost)
+    (name "python-xgboost")
+    (build-system python-build-system)
+    (arguments
+     `(#:phases
+       (modify-phases %standard-phases
+         (add-after 'unpack 'preparations
+           (lambda* (#:key inputs #:allow-other-keys)
+             ;; Move python-package content to parent directory to silence
+             ;; some warnings about files not being found if we chdir.
+             (rename-file "python-package/xgboost" "xgboost")
+             (rename-file "python-package/README.rst" "README.rst")
+             (rename-file "python-package/setup.cfg" "setup.cfg")
+             (rename-file "python-package/setup.py" "setup.py")
+
+             (with-directory-excursion "xgboost"
+               ;; Version is read from a VERSION file, but that file is not
+               ;; part of the installation.
+               (substitute* "__init__.py"
+                 (("__version__ = ''")
+                  (string-append "__version__ = " "'"
+                                 ,(package-version xgboost) "'")))
+               ;; The Python wrapper expects libxgboost to be found in the
+               ;; same source from where the wrapper builds.  This workaround
+               ;; allows libxgboost to be used from the xgboost package.
+               (substitute* "libpath.py"
+                 (("XGBOOST_PATH = ''")
+                  (string-append "XGBOOST_PATH = "
+                                 "'" (assoc-ref inputs "xgboost") "'"))))
+             #t))
+         (replace 'check
+           ;; Python-specific tests are located in tests/python.
+           (lambda* (#:key inputs outputs tests? #:allow-other-keys)
+             (when tests?
+               (add-installed-pythonpath inputs outputs)
+               (invoke "pytest" "tests/python"
+                       ;; FIXME: CLI tests fail with PermissionError.
+                       "--ignore" "tests/python/test_cli.py" "-k"
+                       (string-append
+                        "not test_cli_regression_demo"
+                        ;; The tests below open a network connection.
+                        " and not test_model_compatibility"
+                        " and not test_get_group"
+                        " and not test_cv_no_shuffle"
+                        " and not test_cv"
+                        " and not test_training"
+                        ;; "'['./runexp.sh']' returned non-zero exit status 1"
+                        " and not test_cli_binary_classification")))
+             #t)))))
+    (native-inputs
+     `(("python-pandas" ,python-pandas)
+       ("python-pytest" ,python-pytest)
+       ("python-scikit-learn" ,python-scikit-learn)))
+    (inputs
+     `(("xgboost" ,xgboost)))
+    (propagated-inputs
+     `(("python-numpy" ,python-numpy)
+       ("python-scipy" ,python-scipy)))
+    (synopsis "Python interface for the XGBoost library")))
+
 (define-public python-iml
   (package
     (name "python-iml")
-- 
2.31.1

[bug#48388] [PATCH v3 1/3] gnu: Add dmlc-core.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (dmlc-core): New variable.
---
No changes.

 gnu/packages/machine-learning.scm | 32 +++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 837ebbba67..9d6c5505f7 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1927,6 +1927,38 @@ together building blocks and a subclassing API with an imperative style for
 advanced research.")
     (license license:asl2.0)))
 
+(define-public dmlc-core
+  ;; Version 0.4 is too old for xgboost.
+  (let ((version "0.5")
+        (commit "f00e3ec7abc9f293a1b7061157b0a4e22a735cf5")
+        (revision "0"))
+    (package
+      (name "dmlc-core")
+      (version (git-version version revision commit))
+      (source
+       (origin
+         (method git-fetch)
+         (uri (git-reference
+               (url "https://github.com/dmlc/dmlc-core")
+               (commit commit)))
+         (file-name (git-file-name name version))
+         (sha256
+          (base32 "1x4ad1jhn84fywlk031fmv1kxyiscclmrqn9hhj8gz0mh7z9vcrh"))))
+      (build-system cmake-build-system)
+      (arguments
+       `(#:configure-flags
+         (list "-DGOOGLE_TEST=ON")))
+      (native-inputs
+       `(("googletest" ,googletest)
+         ("python" ,python-wrapper)))
+      (home-page "https://github.com/dmlc/dmlc-core")
+      (synopsis "Common bricks library for machine learning")
+      (description
+       "DMLC-Core is the backbone library to support all DMLC projects,
+offers the bricks to build efficient and scalable distributed machine
+learning libraries.")
+      (license license:asl2.0))))
+
 (define-public python-iml
   (package
     (name "python-iml")
-- 
2.32.0

[bug#48388] [PATCH v3 2/3] gnu: Add xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (xgboost): New variable.
* gnu/packages/patches/xgboost-use-system-dmlc-core.patch: New patch.
* gnu/local.mk (dist_patch_DATA): Add it.
---
Removed the python patch, updated to 1.4.2 and removed "in C++" from synopsis.

 gnu/local.mk                                  |  1 +
 gnu/packages/machine-learning.scm             | 32 +++++++++++++++++
 .../xgboost-use-system-dmlc-core.patch        | 34 +++++++++++++++++++
 3 files changed, 67 insertions(+)
 create mode 100644 gnu/packages/patches/xgboost-use-system-dmlc-core.patch

diff --git a/gnu/local.mk b/gnu/local.mk
index f7ff9af997..19db9787dc 100644
--- a/gnu/local.mk
+++ b/gnu/local.mk
@@ -1842,6 +1842,7 @@ dist_patch_DATA =						\
   %D%/packages/patches/xf86-video-voodoo-pcitag.patch		\
   %D%/packages/patches/xfce4-panel-plugins.patch		\
   %D%/packages/patches/xfce4-settings-defaults.patch		\
+  %D%/packages/patches/xgboost-use-system-dmlc-core.patch       \
   %D%/packages/patches/xmonad-dynamic-linking.patch		\
   %D%/packages/patches/xplanet-1.3.1-cxx11-eof.patch		\
   %D%/packages/patches/xplanet-1.3.1-libdisplay_DisplayOutput.cpp.patch	\
diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 9d6c5505f7..bbdfba960e 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1959,6 +1959,38 @@ offers the bricks to build efficient and scalable distributed machine
 learning libraries.")
       (license license:asl2.0))))
 
+(define-public xgboost
+  (package
+    (name "xgboost")
+    (version "1.4.2")
+    (source
+     (origin
+       (method git-fetch)
+       (uri (git-reference
+             (url "https://github.com/dmlc/xgboost")
+             (commit (string-append "v" version))))
+       (file-name (git-file-name name version))
+       (patches (search-patches "xgboost-use-system-dmlc-core.patch"))
+       (sha256
+        (base32 "00liz816ahk9zj3jv3m2fqwlf6xxfbgvpmpl72iklx32vl192w5d"))))
+    (build-system cmake-build-system)
+    (arguments
+     `(#:configure-flags (list "-DGOOGLE_TEST=ON")))
+    (native-inputs
+     `(("googletest" ,googletest)
+       ("python" ,python-wrapper)))
+    (inputs
+     `(("dmlc-core" ,dmlc-core)))
+    (home-page "https://xgboost.ai/")
+    (synopsis "Gradient boosting (GBDT, GBRT or GBM) library")
+    (description
+     "XGBoost is an optimized distributed gradient boosting library designed
+to be highly efficient, flexible and portable.  It implements machine learning
+algorithms under the Gradient Boosting framework.  XGBoost provides a parallel
+tree boosting (also known as GBDT, GBM) that solve many data science problems
+in a fast and accurate way.")
+    (license license:asl2.0)))
+
 (define-public python-iml
   (package
     (name "python-iml")
diff --git a/gnu/packages/patches/xgboost-use-system-dmlc-core.patch b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
new file mode 100644
index 0000000000..c654694fec
--- /dev/null
+++ b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
@@ -0,0 +1,34 @@
+From 34167cc105e47589b86b6f62da76e0ab744002dd Mon Sep 17 00:00:00 2001
+From: Vinicius Monego <monego@posteo.net>
+Date: Tue, 11 May 2021 19:35:30 -0300
+Subject: [PATCH] Use dmlc-core from Guix.
+
+---
+This patch is a subset of the following Debian patch: https://sources.debian.org/src/xgboost/1.2.1-1/debian/patches/cmake.patch/
+
+Rabit is now developed as part of xgboost, so we don't build it separately.
+
+ CMakeLists.txt | 8 +++-----
+ 1 file changed, 3 insertions(+), 5 deletions(-)
+
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 6cc19fbd..3b42bc17 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -150,11 +150,9 @@ endif (USE_OPENMP)
+
+ # dmlc-core
+ msvc_use_static_runtime()
+-add_subdirectory(${xgboost_SOURCE_DIR}/dmlc-core)
+-set_target_properties(dmlc PROPERTIES
+-  CXX_STANDARD 14
+-  CXX_STANDARD_REQUIRED ON
+-  POSITION_INDEPENDENT_CODE ON)
++add_library(dmlc SHARED IMPORTED)
++find_library(DMLC_LIBRARY dmlc)
++set_property(TARGET dmlc PROPERTY IMPORTED_LOCATION "${DMLC_LIBRARY}")
+ if (MSVC)
+   target_compile_options(dmlc PRIVATE
+                          -D_CRT_SECURE_NO_WARNINGS -D_CRT_SECURE_NO_DEPRECATE)
+--
+2.31.1
-- 
2.32.0

[bug#48388] [PATCH v3 3/3] gnu: Add python-xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (python-xgboost): New variable.
* gnu/packages/patches/python-xgboost-use-system-libxgboost.patch: New patch.
* gnu/local.mk (dist_patch_DATA): Add it.
---
Changed source to inherit from xgboost.
Added phase after 'install to add a VERSION file and symlink libxgboost, so it can be found at runtime. This change allowed to greatly simplify the patch.
Also removed #t from phases.
If this patchset is good enough now, I will do the same changes to LightGBM: http://issues.guix.gnu.org/48785

 gnu/local.mk                                  |  1 +
 gnu/packages/machine-learning.scm             | 66 +++++++++++++++++++
 ...python-xgboost-use-system-libxgboost.patch | 18 +++++
 3 files changed, 85 insertions(+)
 create mode 100644 gnu/packages/patches/python-xgboost-use-system-libxgboost.patch

diff --git a/gnu/local.mk b/gnu/local.mk
index 19db9787dc..821fd390a4 100644
--- a/gnu/local.mk
+++ b/gnu/local.mk
@@ -1632,6 +1632,7 @@ dist_patch_DATA =						\
   %D%/packages/patches/python-unittest2-python3-compat.patch	\
   %D%/packages/patches/python-unittest2-remove-argparse.patch	\
   %D%/packages/patches/python-waitress-fix-tests.patch		\
+  %D%/packages/patches/python-xgboost-use-system-libxgboost.patch \
   %D%/packages/patches/pypy3-7.3.1-fix-tests.patch		\
   %D%/packages/patches/qemu-build-info-manual.patch		\
   %D%/packages/patches/qemu-CVE-2021-20203.patch		\
diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index bbdfba960e..69b96ffd61 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1991,6 +1991,72 @@ tree boosting (also known as GBDT, GBM) that solve many data science problems
 in a fast and accurate way.")
     (license license:asl2.0)))
 
+(define-public python-xgboost
+  (package
+    (inherit xgboost)
+    (name "python-xgboost")
+    (source
+     (origin
+       (inherit (package-source xgboost))
+       (patches (search-patches "python-xgboost-use-system-libxgboost.patch"))))
+    (build-system python-build-system)
+    (arguments
+     `(#:phases
+       (modify-phases %standard-phases
+         (add-after 'unpack 'preparations
+           (lambda* (#:key inputs #:allow-other-keys)
+             ;; Move python-package content to parent directory to silence
+             ;; some warnings about files not being found if we chdir.
+             (rename-file "python-package/xgboost" "xgboost")
+             (rename-file "python-package/README.rst" "README.rst")
+             (rename-file "python-package/setup.cfg" "setup.cfg")
+             (rename-file "python-package/setup.py" "setup.py")))
+         (add-after 'install 'install-version-and-libxgboost
+           (lambda* (#:key inputs outputs #:allow-other-keys)
+             (let* ((out (assoc-ref outputs "out"))
+                    (pylib (string-append out "/lib/python"
+                                          ,(version-major+minor
+                                            (package-version python))
+                                          "/site-packages"))
+                    (xgbdir (string-append pylib "/xgboost"))
+                    (version-file (string-append xgbdir "/VERSION"))
+                    (libxgboost (string-append (assoc-ref inputs "xgboost")
+                                               "/lib/libxgboost.so")))
+               (with-output-to-file version-file
+                 (lambda ()
+                   (display ,(package-version xgboost))))
+               (mkdir-p (string-append xgbdir "/lib"))
+               (symlink libxgboost (string-append xgbdir "/lib"
+                                                  "/libxgboost.so")))))
+         (replace 'check
+           ;; Python-specific tests are located in tests/python.
+           (lambda* (#:key inputs outputs tests? #:allow-other-keys)
+             (when tests?
+               (add-installed-pythonpath inputs outputs)
+               (invoke "pytest" "tests/python"
+                       ;; FIXME: CLI tests fail with PermissionError.
+                       "--ignore" "tests/python/test_cli.py" "-k"
+                       (string-append
+                        "not test_cli_regression_demo"
+                        ;; The tests below open a network connection.
+                        " and not test_model_compatibility"
+                        " and not test_get_group"
+                        " and not test_cv_no_shuffle"
+                        " and not test_cv"
+                        " and not test_training"
+                        ;; "'['./runexp.sh']' returned non-zero exit status 1"
+                        " and not test_cli_binary_classification"))))))))
+    (native-inputs
+     `(("python-pandas" ,python-pandas)
+       ("python-pytest" ,python-pytest)
+       ("python-scikit-learn" ,python-scikit-learn)))
+    (inputs
+     `(("xgboost" ,xgboost)))
+    (propagated-inputs
+     `(("python-numpy" ,python-numpy)
+       ("python-scipy" ,python-scipy)))
+    (synopsis "Python interface for the XGBoost library")))
+
 (define-public python-iml
   (package
     (name "python-iml")
diff --git a/gnu/packages/patches/python-xgboost-use-system-libxgboost.patch b/gnu/packages/patches/python-xgboost-use-system-libxgboost.patch
new file mode 100644
index 0000000000..ce82309afd
--- /dev/null
+++ b/gnu/packages/patches/python-xgboost-use-system-libxgboost.patch
@@ -0,0 +1,18 @@
+This patch was imported from Debian: https://sources.debian.org/src/xgboost/1.2.1-1/debian/patches/setup.py.patch/
+
+Description: the libxgboost.so has been installed by another package already
+Forwarded: no need
+--- a/python-package/setup.py
++++ b/python-package/setup.py
+@@ -286,10 +286,8 @@
+               'numpy',
+               'scipy',
+           ],
+-          ext_modules=[CMakeExtension('libxgboost')],
+           cmdclass={
+               'build_ext': BuildExt,
+               'sdist': Sdist,
+-              'install_lib': InstallLib,
+               'install': Install
+           },
+           extras_require={
\ No newline at end of file
-- 
2.32.0

[bug#48388] [PATCH v4 3/3] gnu: Add python-xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (python-xgboost): New variable.
---
Replaced the patch file with substitutions in 'preparations and changed the lambda arguments to _.

 gnu/packages/machine-learning.scm | 67 +++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 726d491868..b8d79821e0 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -1991,6 +1991,73 @@ tree boosting (also known as GBDT, GBM) that solve many data science problems
 in a fast and accurate way.")
     (license license:asl2.0)))
 
+(define-public python-xgboost
+  (package
+    (inherit xgboost)
+    (name "python-xgboost")
+    (source (package-source xgboost))
+    (build-system python-build-system)
+    (arguments
+     `(#:phases
+       (modify-phases %standard-phases
+         (add-after 'unpack 'preparations
+           (lambda _
+             ;; Move python-package content to parent directory to silence
+             ;; some warnings about files not being found if we chdir.
+             (rename-file "python-package/xgboost" "xgboost")
+             (rename-file "python-package/README.rst" "README.rst")
+             (rename-file "python-package/setup.cfg" "setup.cfg")
+             (rename-file "python-package/setup.py" "setup.py")
+             ;; Skip rebuilding libxgboost.so.
+             (substitute* "setup.py"
+               (("ext_modules=\\[CMakeExtension\\('libxgboost'\\)\\],") "")
+               (("'install_lib': InstallLib,") ""))))
+         (add-after 'install 'install-version-and-libxgboost
+           (lambda* (#:key inputs outputs #:allow-other-keys)
+             (let* ((out (assoc-ref outputs "out"))
+                    (pylib (string-append out "/lib/python"
+                                          ,(version-major+minor
+                                            (package-version python))
+                                          "/site-packages"))
+                    (xgbdir (string-append pylib "/xgboost"))
+                    (version-file (string-append xgbdir "/VERSION"))
+                    (libxgboost (string-append (assoc-ref inputs "xgboost")
+                                               "/lib/libxgboost.so")))
+               (with-output-to-file version-file
+                 (lambda ()
+                   (display ,(package-version xgboost))))
+               (mkdir-p (string-append xgbdir "/lib"))
+               (symlink libxgboost (string-append xgbdir "/lib"
+                                                  "/libxgboost.so")))))
+         (replace 'check
+           ;; Python-specific tests are located in tests/python.
+           (lambda* (#:key inputs outputs tests? #:allow-other-keys)
+             (when tests?
+               (add-installed-pythonpath inputs outputs)
+               (invoke "pytest" "tests/python"
+                       ;; FIXME: CLI tests fail with PermissionError.
+                       "--ignore" "tests/python/test_cli.py" "-k"
+                       (string-append
+                        "not test_cli_regression_demo"
+                        ;; The tests below open a network connection.
+                        " and not test_model_compatibility"
+                        " and not test_get_group"
+                        " and not test_cv_no_shuffle"
+                        " and not test_cv"
+                        " and not test_training"
+                        ;; "'['./runexp.sh']' returned non-zero exit status 1"
+                        " and not test_cli_binary_classification"))))))))
+    (native-inputs
+     `(("python-pandas" ,python-pandas)
+       ("python-pytest" ,python-pytest)
+       ("python-scikit-learn" ,python-scikit-learn)))
+    (inputs
+     `(("xgboost" ,xgboost)))
+    (propagated-inputs
+     `(("python-numpy" ,python-numpy)
+       ("python-scipy" ,python-scipy)))
+    (synopsis "Python interface for the XGBoost library")))
+
 (define-public python-iml
   (package
     (name "python-iml")
-- 
2.32.0

[bug#48388] [PATCH v5 1/3] gnu: Add dmlc-core.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (dmlc-core): New variable.
---
Checked out from v0.5 tag instead of a commit.

 gnu/packages/machine-learning.scm | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 9b5d4c1743..a4768211a0 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -2310,6 +2310,34 @@ learning models.  This package provides the \"lite\" variant for mobile
 devices.")
     (license license:asl2.0)))
 
+(define-public dmlc-core
+  (package
+    (name "dmlc-core")
+    (version "0.5")
+    (source
+     (origin
+       (method git-fetch)
+       (uri (git-reference
+             (url "https://github.com/dmlc/dmlc-core")
+             (commit (string-append "v" version))))
+       (file-name (git-file-name name version))
+       (sha256
+        (base32 "1x4ad1jhn84fywlk031fmv1kxyiscclmrqn9hhj8gz0mh7z9vcrh"))))
+    (build-system cmake-build-system)
+    (arguments
+     `(#:configure-flags
+       (list "-DGOOGLE_TEST=ON")))
+    (native-inputs
+     `(("googletest" ,googletest)
+       ("python" ,python-wrapper)))
+    (home-page "https://github.com/dmlc/dmlc-core")
+    (synopsis "Common bricks library for machine learning")
+    (description
+     "DMLC-Core is the backbone library to support all DMLC projects,
+offers the bricks to build efficient and scalable distributed machine
+learning libraries.")
+    (license license:asl2.0)))
+
 (define-public python-iml
   (package
     (name "python-iml")
-- 
2.30.2

[bug#48388] [PATCH v5 2/3] gnu: Add xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (xgboost): New variable.
* gnu/packages/patches/xgboost-use-system-dmlc-core.patch: New patch.
* gnu/local.mk (dist_patch_DATA): Add it.
---
 gnu/local.mk                                  |  1 +
 gnu/packages/machine-learning.scm             | 32 +++++++++++++++++
 .../xgboost-use-system-dmlc-core.patch        | 34 +++++++++++++++++++
 3 files changed, 67 insertions(+)
 create mode 100644 gnu/packages/patches/xgboost-use-system-dmlc-core.patch

diff --git a/gnu/local.mk b/gnu/local.mk
index 7d706f6f6e..22fc393f0e 100644
--- a/gnu/local.mk
+++ b/gnu/local.mk
@@ -1888,6 +1888,7 @@ dist_patch_DATA =						\
   %D%/packages/patches/xf86-video-voodoo-pcitag.patch		\
   %D%/packages/patches/xfce4-panel-plugins.patch		\
   %D%/packages/patches/xfce4-settings-defaults.patch		\
+  %D%/packages/patches/xgboost-use-system-dmlc-core.patch       \
   %D%/packages/patches/xmonad-dynamic-linking.patch		\
   %D%/packages/patches/xnnpack-system-libraries.patch		\
   %D%/packages/patches/xplanet-1.3.1-cxx11-eof.patch		\
diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index a4768211a0..454088b1a5 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -2338,6 +2338,38 @@ offers the bricks to build efficient and scalable distributed machine
 learning libraries.")
     (license license:asl2.0)))
 
+(define-public xgboost
+  (package
+    (name "xgboost")
+    (version "1.4.2")
+    (source
+     (origin
+       (method git-fetch)
+       (uri (git-reference
+             (url "https://github.com/dmlc/xgboost")
+             (commit (string-append "v" version))))
+       (file-name (git-file-name name version))
+       (patches (search-patches "xgboost-use-system-dmlc-core.patch"))
+       (sha256
+        (base32 "00liz816ahk9zj3jv3m2fqwlf6xxfbgvpmpl72iklx32vl192w5d"))))
+    (build-system cmake-build-system)
+    (arguments
+     `(#:configure-flags (list "-DGOOGLE_TEST=ON")))
+    (native-inputs
+     `(("googletest" ,googletest)
+       ("python" ,python-wrapper)))
+    (inputs
+     `(("dmlc-core" ,dmlc-core)))
+    (home-page "https://xgboost.ai/")
+    (synopsis "Gradient boosting (GBDT, GBRT or GBM) library")
+    (description
+     "XGBoost is an optimized distributed gradient boosting library designed
+to be highly efficient, flexible and portable.  It implements machine learning
+algorithms under the Gradient Boosting framework.  XGBoost provides a parallel
+tree boosting (also known as GBDT, GBM) that solve many data science problems
+in a fast and accurate way.")
+    (license license:asl2.0)))
+
 (define-public python-iml
   (package
     (name "python-iml")
diff --git a/gnu/packages/patches/xgboost-use-system-dmlc-core.patch b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
new file mode 100644
index 0000000000..c654694fec
--- /dev/null
+++ b/gnu/packages/patches/xgboost-use-system-dmlc-core.patch
@@ -0,0 +1,34 @@
+From 34167cc105e47589b86b6f62da76e0ab744002dd Mon Sep 17 00:00:00 2001
+From: Vinicius Monego <monego@posteo.net>
+Date: Tue, 11 May 2021 19:35:30 -0300
+Subject: [PATCH] Use dmlc-core from Guix.
+
+---
+This patch is a subset of the following Debian patch: https://sources.debian.org/src/xgboost/1.2.1-1/debian/patches/cmake.patch/
+
+Rabit is now developed as part of xgboost, so we don't build it separately.
+
+ CMakeLists.txt | 8 +++-----
+ 1 file changed, 3 insertions(+), 5 deletions(-)
+
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 6cc19fbd..3b42bc17 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -150,11 +150,9 @@ endif (USE_OPENMP)
+
+ # dmlc-core
+ msvc_use_static_runtime()
+-add_subdirectory(${xgboost_SOURCE_DIR}/dmlc-core)
+-set_target_properties(dmlc PROPERTIES
+-  CXX_STANDARD 14
+-  CXX_STANDARD_REQUIRED ON
+-  POSITION_INDEPENDENT_CODE ON)
++add_library(dmlc SHARED IMPORTED)
++find_library(DMLC_LIBRARY dmlc)
++set_property(TARGET dmlc PROPERTY IMPORTED_LOCATION "${DMLC_LIBRARY}")
+ if (MSVC)
+   target_compile_options(dmlc PRIVATE
+                          -D_CRT_SECURE_NO_WARNINGS -D_CRT_SECURE_NO_DEPRECATE)
+--
+2.31.1
-- 
2.30.2

[bug#48388] [PATCH v5 3/3] gnu: Add python-xgboost.

[Vinicius Monego]

* gnu/packages/machine-learning.scm (python-xgboost): New variable.
---
Skipped a test that fails in Pandas 1.3.0.

 gnu/packages/machine-learning.scm | 69 +++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 454088b1a5..5e17b09e89 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -2370,6 +2370,75 @@ tree boosting (also known as GBDT, GBM) that solve many data science problems
 in a fast and accurate way.")
     (license license:asl2.0)))
 
+(define-public python-xgboost
+  (package
+    (inherit xgboost)
+    (name "python-xgboost")
+    (source (package-source xgboost))
+    (build-system python-build-system)
+    (arguments
+     `(#:phases
+       (modify-phases %standard-phases
+         (add-after 'unpack 'preparations
+           (lambda _
+             ;; Move python-package content to parent directory to silence
+             ;; some warnings about files not being found if we chdir.
+             (rename-file "python-package/xgboost" "xgboost")
+             (rename-file "python-package/README.rst" "README.rst")
+             (rename-file "python-package/setup.cfg" "setup.cfg")
+             (rename-file "python-package/setup.py" "setup.py")
+             ;; Skip rebuilding libxgboost.so.
+             (substitute* "setup.py"
+               (("ext_modules=\\[CMakeExtension\\('libxgboost'\\)\\],") "")
+               (("'install_lib': InstallLib,") ""))))
+         (add-after 'install 'install-version-and-libxgboost
+           (lambda* (#:key inputs outputs #:allow-other-keys)
+             (let* ((out (assoc-ref outputs "out"))
+                    (pylib (string-append out "/lib/python"
+                                          ,(version-major+minor
+                                            (package-version python))
+                                          "/site-packages"))
+                    (xgbdir (string-append pylib "/xgboost"))
+                    (version-file (string-append xgbdir "/VERSION"))
+                    (libxgboost (string-append (assoc-ref inputs "xgboost")
+                                               "/lib/libxgboost.so")))
+               (with-output-to-file version-file
+                 (lambda ()
+                   (display ,(package-version xgboost))))
+               (mkdir-p (string-append xgbdir "/lib"))
+               (symlink libxgboost (string-append xgbdir "/lib"
+                                                  "/libxgboost.so")))))
+         (replace 'check
+           ;; Python-specific tests are located in tests/python.
+           (lambda* (#:key inputs outputs tests? #:allow-other-keys)
+             (when tests?
+               (add-installed-pythonpath inputs outputs)
+               (invoke "pytest" "tests/python"
+                       ;; FIXME: CLI tests fail with PermissionError.
+                       "--ignore" "tests/python/test_cli.py" "-k"
+                       (string-append
+                        "not test_cli_regression_demo"
+                        ;; The tests below open a network connection.
+                        " and not test_model_compatibility"
+                        " and not test_get_group"
+                        " and not test_cv_no_shuffle"
+                        " and not test_cv"
+                        " and not test_training"
+                        ;; FIXME: May pass in the next version.
+                        " and not test_pandas"
+                        ;; "'['./runexp.sh']' returned non-zero exit status 1"
+                        " and not test_cli_binary_classification"))))))))
+    (native-inputs
+     `(("python-pandas" ,python-pandas)
+       ("python-pytest" ,python-pytest)
+       ("python-scikit-learn" ,python-scikit-learn)))
+    (inputs
+     `(("xgboost" ,xgboost)))
+    (propagated-inputs
+     `(("python-numpy" ,python-numpy)
+       ("python-scipy" ,python-scipy)))
+    (synopsis "Python interface for the XGBoost library")))
+
 (define-public python-iml
   (package
     (name "python-iml")
-- 
2.30.2

bug#48388: [PATCH v5 1/3] gnu: Add dmlc-core.

[Leo Famulari]

On Sat, Sep 25, 2021 at 07:36:49PM +0000, Vinicius Monego wrote:
> * gnu/packages/machine-learning.scm (dmlc-core): New variable.

Thanks! I pushed v5 of this patch series as commit
40a6d8eb7de42248c886c447a299256853054912