topsis tests

skcriteria/tests/test_topsis.py

@@ -0,0 +1,61 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+# License: 3 Clause BSD
+# http://scikit-criteria.org/
+
+
+# =============================================================================
+# FUTURE
+# =============================================================================
+
+from __future__ import unicode_literals
+
+
+# =============================================================================
+# DOC
+# =============================================================================
+
+__doc__ = """test topsis methods"""
+
+
+# =============================================================================
+# IMPORTS
+# =============================================================================
+
+from . import core
+
+from .. import topsis
+from ..common import util
+
+
+# =============================================================================
+# BASE CLASS
+# =============================================================================
+
+class TopsisTest(core.SKCriteriaTestCase):
+
+    def setUp(self):
+        # Data From:
+        # Tzeng, G. H., & Huang, J. J. (2011).
+        # Multiple attribute decision making: methods and applications.
+        # CRC press.
+        self.mtx = [
+            [5, 8, 4],
+            [7, 6, 8],
+            [8, 8, 6],
+            [7, 4, 6],
+        ]
+        self.criteria = [util.MAX, util.MAX, util.MAX]
+
+    def test_topsis(self):
+        weights = [.3, .4, .3]
+
+        result = [3, 2, 1, 4]
+        points = [0.5037, 0.6581, 0.7482, 0.3340]
+
+        rank_result, points_result = topsis.topsis(
+            self.mtx, self.criteria, weights)
+
+        self.assertAllClose(points_result, points, atol=1.e-4)
+        self.assertAllClose(rank_result, result)

skcriteria/topsis.py

@@ -34,7 +34,7 @@ def topsis(mtx, criteria, weights=1):
     nmtx = norm.vector(mtx, axis=0)
 
     # apply weights
-    nweights = norm.vector(weights) if weights is not None else 1
+    nweights = norm.sum(weights) if weights is not None else 1
     wmtx = np.multiply(nmtx, nweights)
 
     # extract mins and maxes