1 ;;; GNU Guix --- Functional package management for GNU
2 ;;; Copyright © 2015, 2016, 2017, 2018, 2019 Ricardo Wurmus <rekado@elephly.net>
3 ;;; Copyright © 2016 Efraim Flashner <efraim@flashner.co.il>
4 ;;; Copyright © 2016, 2017 Marius Bakke <mbakke@fastmail.com>
5 ;;; Copyright © 2016 Hartmut Goebel <h.goebel@crazy-compilers.com>
6 ;;; Copyright © 2018 Tobias Geerinckx-Rice <me@tobias.gr>
7 ;;; Copyright © 2018 Kei Kebreau <kkebreau@posteo.net>
8 ;;; Copyright © 2018 Mark Meyer <mark@ofosos.org>
9 ;;; Copyright © 2018 Ben Woodcroft <donttrustben@gmail.com>
10 ;;; Copyright © 2018 Fis Trivial <ybbs.daans@hotmail.com>
11 ;;; Copyright © 2018 Julien Lepiller <julien@lepiller.eu>
12 ;;; Copyright © 2018 Björn Höfling <bjoern.hoefling@bjoernhoefling.de>
14 ;;; This file is part of GNU Guix.
16 ;;; GNU Guix is free software; you can redistribute it and/or modify it
17 ;;; under the terms of the GNU General Public License as published by
18 ;;; the Free Software Foundation; either version 3 of the License, or (at
19 ;;; your option) any later version.
21 ;;; GNU Guix is distributed in the hope that it will be useful, but
22 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
23 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
24 ;;; GNU General Public License for more details.
26 ;;; You should have received a copy of the GNU General Public License
27 ;;; along with GNU Guix. If not, see <http://www.gnu.org/licenses/>.
29 (define-module (gnu packages machine-learning)
30 #:use-module ((guix licenses) #:prefix license:)
31 #:use-module (guix packages)
32 #:use-module (guix utils)
33 #:use-module (guix download)
34 #:use-module (guix svn-download)
35 #:use-module (guix build-system cmake)
36 #:use-module (guix build-system gnu)
37 #:use-module (guix build-system ocaml)
38 #:use-module (guix build-system python)
39 #:use-module (guix build-system r)
40 #:use-module (guix git-download)
41 #:use-module (gnu packages)
42 #:use-module (gnu packages algebra)
43 #:use-module (gnu packages autotools)
44 #:use-module (gnu packages boost)
45 #:use-module (gnu packages check)
46 #:use-module (gnu packages compression)
47 #:use-module (gnu packages cran)
48 #:use-module (gnu packages dejagnu)
49 #:use-module (gnu packages gcc)
50 #:use-module (gnu packages image)
51 #:use-module (gnu packages maths)
52 #:use-module (gnu packages mpi)
53 #:use-module (gnu packages ocaml)
54 #:use-module (gnu packages onc-rpc)
55 #:use-module (gnu packages perl)
56 #:use-module (gnu packages pkg-config)
57 #:use-module (gnu packages python)
58 #:use-module (gnu packages python-xyz)
59 #:use-module (gnu packages statistics)
60 #:use-module (gnu packages swig)
61 #:use-module (gnu packages xml)
62 #:use-module (gnu packages xorg))
65 ;; The last release is >100 commits behind, so we package from git.
66 (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
69 (version (string-append "2.2.0-1." (string-take commit 8)))
73 (url "https://github.com/libfann/fann.git")
75 (file-name (string-append name "-" version "-checkout"))
78 "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
79 (build-system cmake-build-system)
82 (modify-phases %standard-phases
84 (lambda* (#:key outputs #:allow-other-keys)
85 (let* ((out (assoc-ref outputs "out")))
86 (with-directory-excursion (string-append (getcwd) "/tests")
87 (invoke "./fann_tests"))))))))
88 (home-page "http://leenissen.dk/fann/wp/")
89 (synopsis "Fast Artificial Neural Network")
91 "FANN is a free open source neural network library, which implements
92 multilayer artificial neural networks in C with support for both fully
93 connected and sparsely connected networks.")
94 (license license:lgpl2.1))))
103 (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
104 name "-" version ".tar.gz"))
107 "0zd7s19y5vb7agczl6456bn45cj1y64739sslaskw1qk7dywd0bd"))))
108 (build-system gnu-build-system)
110 `(#:tests? #f ;no "check" target
111 #:phases (modify-phases %standard-phases
114 'install ; no ‘install’ target
115 (lambda* (#:key outputs #:allow-other-keys)
116 (let* ((out (assoc-ref outputs "out"))
117 (bin (string-append out "/bin/")))
119 (for-each (lambda (file)
120 (copy-file file (string-append bin file)))
125 (home-page "http://www.csie.ntu.edu.tw/~cjlin/libsvm/")
126 (synopsis "Library for Support Vector Machines")
128 "LIBSVM is a machine learning library for support vector
129 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
130 distribution estimation (one-class SVM). It supports multi-class
132 (license license:bsd-3)))
134 (define-public python-libsvm
135 (package (inherit libsvm)
136 (name "python-libsvm")
137 (build-system gnu-build-system)
139 `(#:tests? #f ;no "check" target
140 #:make-flags '("-C" "python")
142 (modify-phases %standard-phases
145 'install ; no ‘install’ target
146 (lambda* (#:key inputs outputs #:allow-other-keys)
147 (let ((site (string-append (assoc-ref outputs "out")
151 (assoc-ref inputs "python") 5) 3)
153 (substitute* "python/svm.py"
154 (("../libsvm.so.2") "libsvm.so.2"))
156 (for-each (lambda (file)
157 (copy-file file (string-append site (basename file))))
158 (find-files "python" "\\.py"))
159 (copy-file "libsvm.so.2"
160 (string-append site "libsvm.so.2")))
163 `(("python" ,python)))
164 (synopsis "Python bindings of libSVM")))
167 ;; The latest release candidate is several years and a couple of fixes have
168 ;; been published since. This is why we download the sources from the SVN
170 (let ((svn-revision 2341))
173 (version (string-append "0.9-rc3-0." (number->string svn-revision)))
177 (url "http://svn.code.sf.net/p/ghmm/code/trunk")
178 (revision svn-revision)))
179 (file-name (string-append name "-" version))
182 "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
183 (build-system gnu-build-system)
185 `(#:imported-modules (,@%gnu-build-system-modules
186 (guix build python-build-system))
188 (modify-phases %standard-phases
189 (add-after 'unpack 'enter-dir
190 (lambda _ (chdir "ghmm") #t))
192 (add-after 'install 'check
193 (assoc-ref %standard-phases 'check))
194 (add-before 'check 'fix-PYTHONPATH
195 (lambda* (#:key inputs outputs #:allow-other-keys)
196 (let ((python-version ((@@ (guix build python-build-system)
198 (assoc-ref inputs "python"))))
200 (string-append (getenv "PYTHONPATH")
201 ":" (assoc-ref outputs "out")
202 "/lib/python" python-version
205 (add-after 'enter-dir 'fix-runpath
206 (lambda* (#:key outputs #:allow-other-keys)
207 (substitute* "ghmmwrapper/setup.py"
208 (("^(.*)extra_compile_args = \\[" line indent)
209 (string-append indent
210 "extra_link_args = [\"-Wl,-rpath="
211 (assoc-ref outputs "out") "/lib\"],\n"
214 (assoc-ref outputs "out")
217 (add-after 'enter-dir 'disable-broken-tests
219 (substitute* "tests/Makefile.am"
220 ;; GHMM_SILENT_TESTS is assumed to be a command.
221 (("TESTS_ENVIRONMENT.*") "")
222 ;; Do not build broken tests.
226 (("label_higher_order_test.*$")
227 "label_higher_order_test\n"))
229 ;; These Python unittests are broken as there is no gato.
230 ;; See https://sourceforge.net/p/ghmm/support-requests/3/
231 (substitute* "ghmmwrapper/ghmmunittests.py"
232 (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
234 (string-append indent
235 "@unittest.skip(\"Disabled by Guix\")\n"
238 (add-after 'disable-broken-tests 'autogen
240 (invoke "bash" "autogen.sh"))))))
242 `(("python" ,python-2) ; only Python 2 is supported
243 ("libxml2" ,libxml2)))
245 `(("pkg-config" ,pkg-config)
248 ("autoconf" ,autoconf)
249 ("automake" ,automake)
250 ("libtool" ,libtool)))
251 (home-page "http://ghmm.org")
252 (synopsis "Hidden Markov Model library")
254 "The General Hidden Markov Model library (GHMM) is a C library with
255 additional Python bindings implementing a wide range of types of @dfn{Hidden
256 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
257 training, HMM clustering, HMM mixtures.")
258 (license license:lgpl2.0+))))
267 "http://micans.org/mcl/src/mcl-"
268 (string-replace-substring version "." "-")
272 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
273 (build-system gnu-build-system)
275 `(#:configure-flags (list "--enable-blast")))
278 (home-page "http://micans.org/mcl/")
279 (synopsis "Clustering algorithm for graphs")
281 "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
282 fast and scalable unsupervised cluster algorithm for graphs (also known as
283 networks) based on simulation of (stochastic) flow in graphs.")
284 ;; In the LICENCE file and web page it says "The software is licensed
285 ;; under the GNU General Public License, version 3.", but in several of
286 ;; the source code files it suggests GPL3 or later.
287 ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
288 (license license:gpl3)))
290 (define-public ocaml-mcl
293 (version "12-068oasis4")
298 "https://github.com/fhcrc/mcl/archive/"
300 (file-name (string-append name "-" version ".tar.gz"))
303 "1l5jbhwjpsj38x8b9698hfpkv75h8hn3kj0gihjhn8ym2cwwv110"))))
304 (build-system ocaml-build-system)
306 `(#:ocaml ,ocaml-4.02
307 #:findlib ,ocaml4.02-findlib
309 (modify-phases %standard-phases
310 (add-before 'configure 'patch-paths
312 (substitute* "configure"
313 (("SHELL = /bin/sh") (string-append "SHELL = "(which "sh"))))
314 (substitute* "setup.ml"
316 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh"))))
318 (home-page "https://github.com/fhcrc/mcl")
319 (synopsis "OCaml wrappers around MCL")
321 "This package provides OCaml bindings for the MCL graph clustering
323 (license license:gpl3)))
325 (define-public randomjungle
327 (name "randomjungle")
333 "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
334 "/randomjungle/randomjungle-" version ".tar_.gz"))
335 (patches (search-patches "randomjungle-disable-static-build.patch"))
338 "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
339 (build-system gnu-build-system)
342 (list "--disable-static"
343 (string-append "--with-boost="
344 (assoc-ref %build-inputs "boost")))
346 (modify-phases %standard-phases
348 'configure 'set-CXXFLAGS
350 (setenv "CXXFLAGS" "-fpermissive ")
358 `(("gfortran" ,gfortran)
359 ("gfortran:lib" ,gfortran "lib")))
360 ;; Non-portable assembly instructions are used so building fails on
361 ;; platforms other than x86_64 or i686.
362 (supported-systems '("x86_64-linux" "i686-linux"))
363 (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
364 (synopsis "Implementation of the Random Forests machine learning method")
366 "Random Jungle is an implementation of Random Forests. It is supposed to
367 analyse high dimensional data. In genetics, it can be used for analysing big
368 Genome Wide Association (GWA) data. Random Forests is a powerful machine
369 learning method. Most interesting features are variable selection, missing
370 value imputation, classifier creation, generalization error estimation and
371 sample proximities between pairs of cases.")
372 (license license:gpl3+)))
374 (define-public openfst
380 (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
381 "FstDownload/openfst-" version ".tar.gz"))
384 "0x9wfcqd8hq4h349s7j77sr60h8xjdfshqw1m3a2n6z5bdr9qkm1"))))
385 (build-system gnu-build-system)
386 (home-page "http://www.openfst.org")
387 (synopsis "Library for weighted finite-state transducers")
388 (description "OpenFst is a library for constructing, combining,
389 optimizing, and searching weighted finite-state transducers (FSTs).")
390 (license license:asl2.0)))
392 (define-public shogun
400 "ftp://shogun-toolbox.org/shogun/releases/"
401 (version-major+minor version)
402 "/sources/shogun-" version ".tar.bz2"))
405 "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
406 (modules '((guix build utils)
410 ;; Remove non-free sources and files referencing them
411 (for-each delete-file
412 (find-files "src/shogun/classifier/svm/"
413 "SVMLight\\.(cpp|h)"))
414 (for-each delete-file
415 (find-files "examples/undocumented/libshogun/"
417 "(classifier_.*svmlight.*|"
418 "evaluation_cross_validation_locked_comparison).cpp")))
419 ;; Remove non-free functions.
420 (define (delete-ifdefs file)
421 (with-atomic-file-replacement file
423 (let loop ((line (read-line in 'concat))
425 (if (eof-object? line)
430 "#endif //USE_SVMLIGHT" line)))
432 "#ifdef USE_SVMLIGHT" line))))
433 (when (or (not skipping?)
434 (and skipping? (not skip-next?)))
436 (loop (read-line in 'concat) skip-next?)))))))
437 (for-each delete-ifdefs
439 (find-files "src/shogun/classifier/mkl"
440 "^MKLClassification\\.cpp")
441 (find-files "src/shogun/classifier/svm"
442 "^SVMLightOneClass\\.(cpp|h)")
443 (find-files "src/shogun/multiclass"
444 "^ScatterSVM\\.(cpp|h)")
445 (find-files "src/shogun/kernel/"
446 "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
447 (find-files "src/shogun/regression/svr"
448 "^(MKLRegression|SVRLight)\\.(cpp|h)")
449 (find-files "src/shogun/transfer/domain_adaptation"
450 "^DomainAdaptationSVM\\.(cpp|h)")))
452 (build-system cmake-build-system)
454 '(#:tests? #f ;no check target
456 (modify-phases %standard-phases
457 (add-after 'unpack 'delete-broken-symlinks
459 (for-each delete-file '("applications/arts/data"
460 "applications/asp/data"
461 "applications/easysvm/data"
462 "applications/msplicer/data"
463 "applications/ocr/data"
465 "examples/undocumented/data"))
467 (add-after 'unpack 'change-R-target-path
468 (lambda* (#:key outputs #:allow-other-keys)
469 (substitute* '("src/interfaces/r/CMakeLists.txt"
470 "examples/meta/r/CMakeLists.txt")
471 (("\\$\\{R_COMPONENT_LIB_PATH\\}")
472 (string-append (assoc-ref outputs "out")
475 (add-after 'unpack 'fix-octave-modules
476 (lambda* (#:key outputs #:allow-other-keys)
477 (substitute* "src/interfaces/octave/CMakeLists.txt"
478 (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
479 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
480 ;; change target directory
481 (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
482 (string-append (assoc-ref outputs "out")
483 "/share/octave/packages")))
484 (substitute* '("src/interfaces/octave/swig_typemaps.i"
485 "src/interfaces/octave/sg_print_functions.cpp")
486 ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
487 (("octave/config\\.h") "octave/octave-config.h")
488 (("octave/oct-obj.h") "octave/ovl.h"))
490 (add-after 'unpack 'move-rxcpp
491 (lambda* (#:key inputs #:allow-other-keys)
492 (let ((rxcpp-dir "shogun/third-party/rxcpp"))
494 (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
496 (add-before 'build 'set-HOME
497 ;; $HOME needs to be set at some point during the build phase
498 (lambda _ (setenv "HOME" "/tmp") #t)))
500 (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
501 "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
502 "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
503 ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
504 ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
505 ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
506 ;;"-DINTERFACE_LUA=ON" ;fails because lua doesn't build pkgconfig file
507 "-DINTERFACE_OCTAVE=ON"
508 "-DINTERFACE_PYTHON=ON"
509 "-DINTERFACE_R=ON")))
512 ("numpy" ,python-numpy)
513 ("r-minimal" ,r-minimal)
514 ("octave" ,octave-cli)
519 ("arpack" ,arpack-ng)
526 `(("pkg-config" ,pkg-config)
528 ;; Non-portable SSE instructions are used so building fails on platforms
529 ;; other than x86_64.
530 (supported-systems '("x86_64-linux"))
531 (home-page "http://shogun-toolbox.org/")
532 (synopsis "Machine learning toolbox")
534 "The Shogun Machine learning toolbox provides a wide range of unified and
535 efficient Machine Learning (ML) methods. The toolbox seamlessly allows to
536 combine multiple data representations, algorithm classes, and general purpose
537 tools. This enables both rapid prototyping of data pipelines and extensibility
538 in terms of new algorithms.")
539 (license license:gpl3+)))
548 (uri (string-append "https://github.com/ReactiveX/RxCpp/archive/v"
552 "0y2isr8dy2n1yjr9c5570kpc9lvdlch6jv0jvw000amwn5d3krsh"))
553 (file-name (string-append name "-" version ".tar.gz"))))
554 (build-system cmake-build-system)
557 (modify-phases %standard-phases
558 (add-after 'unpack 'remove-werror
560 (substitute* (find-files ".")
565 (invoke "ctest"))))))
567 `(("catch" ,catch-framework)))
568 (home-page "http://reactivex.io/")
569 (synopsis "Reactive Extensions for C++")
571 "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
572 values-distributed-in-time. ReactiveX is a library for composing asynchronous
573 and event-based programs by using observable sequences.
575 It extends the observer pattern to support sequences of data and/or events and
576 adds operators that allow you to compose sequences together declaratively while
577 abstracting away concerns about things like low-level threading,
578 synchronization, thread-safety, concurrent data structures, and non-blocking
580 (license license:asl2.0)))
582 (define-public r-adaptivesparsity
584 (name "r-adaptivesparsity")
588 (uri (cran-uri "AdaptiveSparsity" version))
591 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
593 `((upstream-name . "AdaptiveSparsity")))
594 (build-system r-build-system)
597 (modify-phases %standard-phases
598 (add-after 'unpack 'link-against-armadillo
600 (substitute* "src/Makevars"
601 (("PKG_LIBS=" prefix)
602 (string-append prefix "-larmadillo"))))))))
605 ("r-matrix" ,r-matrix)
607 ("r-rcpparmadillo" ,r-rcpparmadillo)))
609 `(("armadillo" ,armadillo)))
610 (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
611 (synopsis "Adaptive sparsity models")
613 "This package implements the Figueiredo machine learning algorithm for
614 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
616 (license license:lgpl3+)))
618 (define-public r-kernlab
625 (uri (cran-uri "kernlab" version))
628 "1m0xqf6gyvwayz7w3c83y32ayvnlz0jicj8ijk808zq9sh7dbbgn"))))
629 (build-system r-build-system)
630 (home-page "https://cran.r-project.org/web/packages/kernlab")
631 (synopsis "Kernel-based machine learning tools")
633 "This package provides kernel-based machine learning methods for
634 classification, regression, clustering, novelty detection, quantile regression
635 and dimensionality reduction. Among other methods @code{kernlab} includes
636 Support Vector Machines, Spectral Clustering, Kernel PCA, Gaussian Processes
638 (license license:gpl2)))
647 "http://dlib.net/files/dlib-" version ".tar.bz2"))
650 "1mljz02kwkrbggyncxv5fpnyjdybw2qihaacb3js8yfkw12vwpc2"))
651 (modules '((guix build utils)))
654 ;; Delete ~13MB of bundled dependencies.
655 (delete-file-recursively "dlib/external")
656 (delete-file-recursively "docs/dlib/external")
658 (build-system cmake-build-system)
661 (modify-phases %standard-phases
662 (add-after 'unpack 'disable-asserts
664 ;; config.h recommends explicitly enabling or disabling asserts
665 ;; when building as a shared library. By default neither is set.
666 (substitute* "dlib/config.h"
667 (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
669 (add-after 'disable-asserts 'disable-failing-tests
671 ;; One test times out on MIPS, so we need to disable it.
672 ;; Others are flaky on some platforms.
673 (let* ((system ,(or (%current-target-system)
675 (disabled-tests (cond
676 ((string-prefix? "mips64" system)
677 '("object_detector" ; timeout
679 ((string-prefix? "armhf" system)
680 '("learning_to_track"))
681 ((string-prefix? "i686" system)
686 (substitute* "dlib/test/makefile"
687 (((string-append "SRC \\+= " test "\\.cpp")) "")))
692 ;; No test target, so we build and run the unit tests here.
693 (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
694 (with-directory-excursion test-dir
695 (invoke "make" "-j" (number->string (parallel-job-count)))
696 (invoke "./dtest" "--runall"))
698 (add-after 'install 'delete-static-library
699 (lambda* (#:key outputs #:allow-other-keys)
700 (delete-file (string-append (assoc-ref outputs "out")
704 `(("pkg-config" ,pkg-config)
713 ("openblas" ,openblas)
716 "Toolkit for making machine learning and data analysis applications in C++")
718 "Dlib is a modern C++ toolkit containing machine learning algorithms and
719 tools. It is used in both industry and academia in a wide range of domains
720 including robotics, embedded devices, mobile phones, and large high performance
721 computing environments.")
722 (home-page "http://dlib.net")
723 (license license:boost1.0)))
725 (define-public python-scikit-learn
727 (name "python-scikit-learn")
733 (url "https://github.com/scikit-learn/scikit-learn.git")
735 (file-name (git-file-name name version))
738 "0fkhwg3xn1s7ln9q1szq6kwc4jhwvjh8w4kmv9wcrqy7cq3lbv0d"))))
739 (build-system python-build-system)
742 (modify-phases %standard-phases
743 (add-after 'build 'build-ext
744 (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
747 ;; Restrict OpenBLAS threads to prevent segfaults while testing!
748 (setenv "OPENBLAS_NUM_THREADS" "1")
750 ;; Some tests require write access to $HOME.
751 (setenv "HOME" "/tmp")
753 (invoke "pytest" "sklearn" "-m" "not network")))
754 ;; FIXME: This fails with permission denied
755 (delete 'reset-gzip-timestamps))))
757 `(("openblas" ,openblas)))
759 `(("python-pytest" ,python-pytest)
760 ("python-pandas" ,python-pandas) ;for tests
761 ("python-cython" ,python-cython)))
763 `(("python-numpy" ,python-numpy)
764 ("python-scipy" ,python-scipy)))
765 (home-page "http://scikit-learn.org/")
766 (synopsis "Machine Learning in Python")
768 "Scikit-learn provides simple and efficient tools for data mining and
770 (license license:bsd-3)))
772 (define-public python2-scikit-learn
773 (package-with-python2 python-scikit-learn))
775 (define-public python-autograd
776 (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
778 (version (git-version "0.0.0" revision commit)))
780 (name "python-autograd")
781 (home-page "https://github.com/HIPS/autograd")
789 "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
790 (file-name (git-file-name name version))))
792 (build-system python-build-system)
794 `(("python-nose" ,python-nose)
795 ("python-pytest" ,python-pytest)))
797 `(("python-future" ,python-future)
798 ("python-numpy" ,python-numpy)))
800 `(#:phases (modify-phases %standard-phases
803 (invoke "py.test" "-v"))))))
804 (synopsis "Efficiently computes derivatives of NumPy code")
805 (description "Autograd can automatically differentiate native Python and
806 NumPy code. It can handle a large subset of Python's features, including loops,
807 ifs, recursion and closures, and it can even take derivatives of derivatives
808 of derivatives. It supports reverse-mode differentiation
809 (a.k.a. backpropagation), which means it can efficiently take gradients of
810 scalar-valued functions with respect to array-valued arguments, as well as
811 forward-mode differentiation, and the two can be composed arbitrarily. The
812 main intended application of Autograd is gradient-based optimization.")
813 (license license:expat))))
815 (define-public python2-autograd
816 (package-with-python2 python-autograd))
818 (define-public lightgbm
825 "https://github.com/Microsoft/LightGBM/archive/v"
829 "132zf0yk0545mg72hyzxm102g3hpb6ixx9hnf8zd2k55gas6cjj1"))
830 (file-name (string-append name "-" version ".tar.gz"))))
832 `(("python-pytest" ,python-pytest)
833 ("python-nose" ,python-nose)))
835 `(("openmpi" ,openmpi)))
837 `(("python-numpy" ,python-numpy)
838 ("python-scipy" ,python-scipy)))
843 (modify-phases %standard-phases
845 (lambda* (#:key outputs #:allow-other-keys)
846 (with-directory-excursion ,(string-append "../LightGBM-" version)
847 (invoke "pytest" "tests/c_api_test/test_.py")))))))
848 (build-system cmake-build-system)
849 (home-page "https://github.com/Microsoft/LightGBM")
850 (synopsis "Gradient boosting framework based on decision tree algorithms")
851 (description "LightGBM is a gradient boosting framework that uses tree
852 based learning algorithms. It is designed to be distributed and efficient with
853 the following advantages:
856 @item Faster training speed and higher efficiency
857 @item Lower memory usage
858 @item Better accuracy
859 @item Parallel and GPU learning supported (not enabled in this package)
860 @item Capable of handling large-scale data
862 (license license:expat)))
864 (define-public vowpal-wabbit
865 ;; Language bindings not included.
867 (name "vowpal-wabbit")
872 "https://github.com/JohnLangford/vowpal_wabbit/archive/"
876 "0clp2kb7rk5sckhllxjr5a651awf4s8dgzg4659yh4hf5cqnf0gr"))
877 (file-name (string-append name "-" version ".tar.gz"))))
883 (list (string-append "--with-boost="
884 (assoc-ref %build-inputs "boost")))))
885 (build-system gnu-build-system)
886 (home-page "https://github.com/JohnLangford/vowpal_wabbit")
887 (synopsis "Fast machine learning library for online learning")
888 (description "Vowpal Wabbit is a machine learning system with techniques
889 such as online, hashing, allreduce, reductions, learning2search, active, and
890 interactive learning.")
891 (license license:bsd-3)))
893 (define-public python2-fastlmm
895 (name "python2-fastlmm")
900 (uri (pypi-uri "fastlmm" version ".zip"))
903 "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
904 (build-system python-build-system)
906 `(#:tests? #f ; some test files are missing
907 #:python ,python-2)) ; only Python 2.7 is supported
909 `(("python2-numpy" ,python2-numpy)
910 ("python2-scipy" ,python2-scipy)
911 ("python2-matplotlib" ,python2-matplotlib)
912 ("python2-pandas" ,python2-pandas)
913 ("python2-scikit-learn" ,python2-scikit-learn)
914 ("python2-pysnptools" ,python2-pysnptools)))
917 ("python2-cython" ,python2-cython)
918 ("python2-mock" ,python2-mock)
919 ("python2-nose" ,python2-nose)))
920 (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
921 (synopsis "Perform genome-wide association studies on large data sets")
923 "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
924 Models, is a program for performing both single-SNP and SNP-set genome-wide
925 association studies (GWAS) on extremely large data sets.")
926 (license license:asl2.0)))