Example #1
0
 def test_idf_5(self):
     dtc = [ ["a", "b", "a", "c"], 
             ["b", "d"] ]
     w = idf(dtc)
     self.assertEqual(w, (["a", "b", "c", "d"],
                          [[log(2/1.0), # a
                            log(2/2.0), # b 
                            log(2/1.0), # c
                            log(2/1.0)  # d
                            ], 
                            [log(2/1.0), # a
                             log(2/2.0), # b 
                             log(2/1.0), # c
                             log(2/1.0)  # d
                         ] 
                       ]))
Example #2
0
 def test_idf_normalized(self):
     dtc = [ ["a", "b", "a", "c"], 
             ["b", "d"] ]
     w = idf(dtc, normalized=True)
     self.assertEqual(w, (["a", "b", "c", "d"],
                          [[log(2/1.0) / log(2), # a
                            log(2/2.0) / log(2), # b 
                            log(2/1.0) / log(2), # c
                            log(2/1.0) / log(2)  # d
                            ], 
                            [log(2/1.0) / log(2), # a
                             log(2/2.0) / log(2), # b 
                             log(2/1.0) / log(2), # c
                             log(2/1.0) / log(2)  # d
                         ] 
                       ]))    
Example #3
0
 def test_idf_5(self):
     dtc = [["a", "b", "a", "c"], ["b", "d"]]
     w = idf(dtc)
     self.assertEqual(
         w,
         (
             ["a", "b", "c", "d"],
             [
                 [
                     log(2 / 1.0),  # a
                     log(2 / 2.0),  # b 
                     log(2 / 1.0),  # c
                     log(2 / 1.0)  # d
                 ],
                 [
                     log(2 / 1.0),  # a
                     log(2 / 2.0),  # b 
                     log(2 / 1.0),  # c
                     log(2 / 1.0)  # d
                 ]
             ]))
Example #4
0
 def test_idf_normalized(self):
     dtc = [["a", "b", "a", "c"], ["b", "d"]]
     w = idf(dtc, normalized=True)
     self.assertEqual(
         w,
         (
             ["a", "b", "c", "d"],
             [
                 [
                     log(2 / 1.0) / log(2),  # a
                     log(2 / 2.0) / log(2),  # b 
                     log(2 / 1.0) / log(2),  # c
                     log(2 / 1.0) / log(2)  # d
                 ],
                 [
                     log(2 / 1.0) / log(2),  # a
                     log(2 / 2.0) / log(2),  # b 
                     log(2 / 1.0) / log(2),  # c
                     log(2 / 1.0) / log(2)  # d
                 ]
             ]))
Example #5
0
 def test_idf_4(self):
     dtc = [["a"], ["b"]]
     w = idf(dtc)
     self.assertEqual(w, (["a", "b"], [[log(
         2 / 1.0), log(2 / 1.0)], [log(2 / 1.0), log(2 / 1.0)]]))
Example #6
0
 def test_idf_3(self):
     dtc = [["a"], ["a"]]
     w = idf(dtc)
     self.assertEqual(w, (["a"], [[log(2 / 2.0)], [log(2 / 2.0)]]))
Example #7
0
 def test_idf_1(self):
     dtc = [["a"]]
     w = idf(dtc)
     self.assertEqual(w, (["a"], [[log(1 / 1.0)]]))
Example #8
0
 def test_idf_empty_3(self):
     dtc = [["a"], []]
     self.assertEqual(idf(dtc), (["a"], [[log(2 / 1)], [log(2 / 1.0)]]))
Example #9
0
 def test_idf_empty_2(self):
     dtc = [[]]
     self.assertEqual(idf(dtc), ([], [[]]))
Example #10
0
 def test_idf_empty_1(self):
     dtc = []
     self.assertEqual(idf(dtc), ([], []))
Example #11
0
 def test_idf_4(self):
     dtc = [ ["a"], ["b"] ]
     w = idf(dtc)
     self.assertEqual(w, (["a", "b"],
                          [[log(2/1.0), log(2/1.0)], 
                           [log(2/1.0), log(2/1.0)]]))
Example #12
0
 def test_idf_3(self):
     dtc = [ ["a"], ["a"] ]
     w = idf(dtc)
     self.assertEqual(w, (["a"], 
                          [[log(2/2.0)], 
                           [log(2/2.0)]]))
Example #13
0
 def test_idf_1(self):
     dtc = [ ["a"] ]
     w = idf(dtc)
     self.assertEqual(w, (["a"], 
                          [[log(1/1.0)]]))
Example #14
0
 def test_idf_empty_3(self):
     dtc = [["a"], []]
     self.assertEqual(idf(dtc), (["a"], [[log(2/1)], [log(2/1.0)]]))
Example #15
0
 def test_idf_empty_2(self):
     dtc = [[]]
     self.assertEqual(idf(dtc), ([], [[]]))
Example #16
0
 def test_idf_empty_1(self):
     dtc = []
     self.assertEqual(idf(dtc), ([], []))