misc/py-shap2: New port: Unified approach to explain the output of any machine learning model

misc/Makefile

@@ -437,6 +437,7 @@
     SUBDIR += py-qiskit-machine-learning
     SUBDIR += py-scikit-fusion
     SUBDIR += py-serverfiles
+    SUBDIR += py-shap2
     SUBDIR += py-soapy_power
     SUBDIR += py-spdx
     SUBDIR += py-spdx-lookup

misc/py-shap2/Makefile

@@ -0,0 +1,30 @@
+PORTNAME=	shap2
+DISTVERSION=	0.42.0
+CATEGORIES=	misc # machine-learning
+MASTER_SITES=	CHEESESHOP
+PKGNAMEPREFIX=	${PYTHON_PKGNAMEPREFIX}
+
+MAINTAINER=	yuri@FreeBSD.org
+COMMENT=	Unified approach to explain the output of any machine learning model
+WWW=		https://github.com/slundberg/shap
+
+LICENSE=	MIT
+LICENSE_FILE=	${WRKSRC}/LICENSE
+
+PY_DEPENDS=	${PYNUMPY} \
+		${PYTHON_PKGNAMEPREFIX}numba>0:devel/py-numba@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}pandas>0:math/py-pandas@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}requests>0:www/py-requests@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}scikit-learn>0:science/py-scikit-learn@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}scipy>0:science/py-scipy@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}xgboost>0:misc/py-xgboost@${PY_FLAVOR}
+BUILD_DEPENDS=	${PY_DEPENDS}
+RUN_DEPENDS=	${PY_DEPENDS}
+
+USES=		python:3.6+
+USE_PYTHON=	distutils autoplist
+
+post-install:
+	@${STRIP_CMD} ${STAGEDIR}${PYTHON_SITELIBDIR}/shap2/_cext${PYTHON_EXT_SUFFIX}.so
+
+.include <bsd.port.mk>

misc/py-shap2/distinfo

@@ -0,0 +1,3 @@
+TIMESTAMP = 1673059649
+SHA256 (shap2-0.42.0.tar.gz) = a425c963131ea3e1f6c217e524821f9c2ae4d2e97d8ad5fd33f80fd175df5960
+SIZE (shap2-0.42.0.tar.gz) = 377323

misc/py-shap2/files/patch-shap2_cext_tree__shap.h

@@ -0,0 +1,11 @@
+--- shap2/cext/tree_shap.h.orig	2023-01-07 02:47:51 UTC
++++ shap2/cext/tree_shap.h
+@@ -13,7 +13,7 @@
+ #include <ctime>
+ #if defined(_WIN32) || defined(WIN32)
+     #include <malloc.h>
+-#elif defined(__MVS__)
++#elif defined(__MVS__) || defined(__FreeBSD__)
+     #include <stdlib.h>
+ #else
+     #include <alloca.h>

misc/py-shap2/pkg-descr

@@ -0,0 +1,4 @@
+SHAP (SHapley Additive exPlanations) is a game theoretic approach to explain
+the output of any machine learning model. It connects optimal credit
+allocation with local explanations using the classic Shapley values from game
+theory and their related extensions.