remove adjustOccs from Candidate

scikit-mine · Mar 20, 2023 · e5188e1 · e5188e1
1 parent 50865d3
commit e5188e1
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 52 deletions.
diff --git a/skmine/periodic/candidate.py b/skmine/periodic/candidate.py
@@ -289,23 +289,6 @@ def getCostUncoveredRatio(self):
     def isEfficient(self, dcosts):
         return (self.getCost() / self.getNbUncovered()) < np.mean([dcosts[unc[1]] for unc in self.uncov])
 
-    def adjustOccs(self):
-        """
-        Adds to self.P["occs_up"], the list of uncov timestamps
-
-        Returns
-        -------
-        None
-        """
-        if not self.isPattern() and (self.uncov is not None) and (self.getNbUncovered() < self.getNbOccs()):
-            okk = self.getEvOccs()
-            mni, mxi = (0, len(self.O) - 1)
-            while okk[mni] not in self.uncov:
-                mni += 1
-            while okk[mxi] not in self.uncov:
-                mxi -= 1
-            self.P["occs_up"] = [self.O[kk] for kk in range(mni, mxi + 1)]
-
     def getProps(self, nkey=0, max_offset=None):
         if max_offset is None:
             max_offset = PROPS_MAX_OFFSET

diff --git a/skmine/periodic/run_mine.py b/skmine/periodic/run_mine.py
@@ -732,15 +732,14 @@ def makeCandOnOrder(cand_pids, data_details, patterns_props, cpool):
     return new_cand
 
 
-def filter_candidates_cover(cands, dcosts, min_cov=1, adjust_occs=False, cis=None):
+def filter_candidates_cover(cands, dcosts, min_cov=1, cis=None):
     """
     Filters a list of candidates based on their coverage and cost efficiency.
 
     Parameters:
     cands (dict or list): A dictionary or list of Candidate objects.
     dcosts (dict): A dictionary of costs for each data point.
     min_cov (int): The minimum number of data points a candidate must cover to be selected. Default is 1.
-    adjust_occs (bool): Whether to adjust occurrences of patterns. Default is False.
     cis (list): A list of candidate indices to consider. If not provided, all candidates will be considered.
 
     Returns:
@@ -750,7 +749,7 @@ def filter_candidates_cover(cands, dcosts, min_cov=1, adjust_occs=False, cis=Non
     ------
     >>> cands = [Candidate(...), Candidate(...), ...]
     >>> dcosts = {'data1': 0.5, 'data2': 0.8, ...}
-    >>> selected = filter_candidates_cover(cands, dcosts, min_cov=2, adjust_occs=True, cis=[0, 2, 5])
+    >>> selected = filter_candidates_cover(cands, dcosts, min_cov=2,  cis=[0, 2, 5])
     """
     if cis is None:
         if type(cands) is dict:
@@ -771,8 +770,6 @@ def filter_candidates_cover(cands, dcosts, min_cov=1, adjust_occs=False, cis=Non
         nxti = cis.pop(0)
         if cands[nxti].getCostUncoveredRatio() <= max_eff:
             if cands[nxti].getNbUncovered() >= min_cov and cands[nxti].isEfficient(dcosts):
-                if not cands[nxti].isPattern() and adjust_occs:
-                    cands[nxti].adjustOccs()
 
                 selected.append(nxti)
                 covered.update(cands[nxti].getUncovered())
@@ -1013,7 +1010,7 @@ def mine_seqs(seqs, complex=True, max_p=None):
     results["Final_selection_TIME"] = str(tac_comb)
 
     selected = filter_candidates_cover(
-        cdict, dcosts, min_cov=3, adjust_occs=True)
+        cdict, dcosts, min_cov=3)
     pc = PatternCollection([cdict[c].getPattT0E() for c in selected])
     tac = datetime.datetime.now()
 

diff --git a/skmine/periodic/tests/test_candidate.py b/skmine/periodic/tests/test_candidate.py
@@ -147,42 +147,13 @@ def test___str__(cand):
         50.22386061518789, 3, cand.getCostRatio(), 3, {'alpha': 1, 'p': 50, 'source': (2, 0)}, 0)
 
 
-def test_adjustOccs_no_changes(cand, cand_pattern):
-    P = cand.P
-    cand.adjustOccs()
-    assert cand.P == P
-
-    P = cand_pattern.P
-    cand_pattern.adjustOccs()
-    assert cand_pattern.P == P
-
-
 def test_initUncovered(cand, cand_pattern):
     cand.initUncovered()
     assert cand.uncov == {(0, 1), (51, 1), (100, 1)}
     cand_pattern.initUncovered()
     assert cand_pattern.uncov == {(0, 4), (60480, 4), (120961, 4), (181441, 4), (241921, 4)}
 
 
-def test_adjustOccs():
-    cid = 1
-    pattern = {
-        'alpha': 10,
-        'p': 75,
-        'pos': [8, 9, 11, 12, 14, 15, 16],
-        'uncov': {8, 9, 11, 12, 14, 15, 16},
-        'source': (1, 1)
-    }
-    O = [159626778, 159626862, 159626934, 159627012, 159627090, 159627162, 159627234]
-    E = [9, -3, 3, 3, -3, -3]
-    cost = 78.81334894354123
-    cand = Candidate(cid, pattern, O, E, cost)
-    cand.uncov = {(159626862, 10), (159626934, 10), (159627012, 10), (159627090, 10), (159627162, 10), (159627234, 10)}
-    # 6 uncovered and 7 occurrences
-    cand.adjustOccs()
-    assert cand.P["occs_up"] == [159626862, 159626934, 159627012, 159627090, 159627162, 159627234]
-
-
 def test_factorizePattern():
     cid = 1
     pattern = Pattern(