def update_for_cols(membership, cd_scores, multiprocessing):
"""updating column memberships according to cd_scores"""
global UPDATE_MEMBERSHIP
colnames = cd_scores.row_names
best_clusters = get_best_clusters(cd_scores,
membership.num_clusters_per_column())
max_changes = membership.max_changes_per_col()
change_prob = membership.probability_seeing_col_change()
for index in xrange(cd_scores.num_rows):
col = colnames[index]
clusters = best_clusters[col]
if seeing_change(change_prob):
for c in range(max_changes):
if len(clusters) > 0:
free_slots = membership.free_slots_for_column(col)
if len(free_slots) > 0:
slot = free_slots[0]
"""
the slot can be out of bounds for the clusters array when
the setting for clusters_per_row/clusters_per_col is too
large, in this case pick a spot inside the array to avoid
the exception"""
if slot > len(clusters) - 1:
slot = len(clusters) - 1
take_cluster = clusters[slot]
#print "ii = ", c, ", add cluster ", take_cluster, " at ", free_slots[0]
membership.add_cluster_to_column(col, take_cluster)
else:
col_clusters = membership.col_membs[
membership.colidx[col]]
multi = util.which_multiple(col_clusters)
if len(multi) > 0:
# indexes of col_clusters that are in multiple
for i, cluster in enumerate(col_clusters):
if cluster in multi:
#print "multiple in row: ", index, " ii: ", c, " col.change ", i, " -> ", clusters[i]
membership.replace_column_cluster(
col, i, clusters[i])
break
else:
replace_delta_column_member(
membership, col, clusters, cd_scores)
def update_for_cols(membership, cd_scores, multiprocessing):
"""updating column memberships according to cd_scores"""
global UPDATE_MEMBERSHIP
colnames = cd_scores.row_names
best_clusters = get_best_clusters(cd_scores, membership.num_clusters_per_column())
max_changes = membership.max_changes_per_col()
change_prob = membership.probability_seeing_col_change()
for index in xrange(cd_scores.num_rows):
col = colnames[index]
clusters = best_clusters[col]
if seeing_change(change_prob):
for c in range(max_changes):
if len(clusters) > 0:
free_slots = membership.free_slots_for_column(col)
if len(free_slots) > 0:
slot = free_slots[0]
"""
the slot can be out of bounds for the clusters array when
the setting for clusters_per_row/clusters_per_col is too
large, in this case pick a spot inside the array to avoid
the exception"""
if slot > len(clusters) - 1:
slot = len(clusters) - 1
take_cluster = clusters[slot]
#print "ii = ", c, ", add cluster ", take_cluster, " at ", free_slots[0]
membership.add_cluster_to_column(col, take_cluster)
else:
col_clusters = membership.col_membs[membership.colidx[col]]
multi = util.which_multiple(col_clusters)
if len(multi) > 0:
# indexes of col_clusters that are in multiple
for i, cluster in enumerate(col_clusters):
if cluster in multi:
#print "multiple in row: ", index, " ii: ", c, " col.change ", i, " -> ", clusters[i]
membership.replace_column_cluster(col, i, clusters[i])
break
else:
replace_delta_column_member(membership, col, clusters, cd_scores)
def test_which_multiple_some_multiple(self):
"""tests the which_multiple() function"""
multiple = util.which_multiple([1, 2, 3, 1, 2])
self.assertEquals(2, len(multiple))
self.assertTrue(1 in multiple)
self.assertTrue(2 in multiple)
def test_which_multiple_all_unique(self):
"""tests the which_multiple() function"""
self.assertEquals(0, len(util.which_multiple([1, 2, 3])))
def test_which_multiple_none(self):
"""tests the which_multiple() function"""
self.assertEquals(0, len(util.which_multiple([])))
def test_which_multiple_some_multiple(self):
"""tests the which_multiple() function"""
multiple = util.which_multiple([1, 2, 3, 1, 2])
self.assertEquals(2, len(multiple))
self.assertTrue(1 in multiple)
self.assertTrue(2 in multiple)
def test_which_multiple_all_unique(self):
"""tests the which_multiple() function"""
self.assertEquals(0, len(util.which_multiple([1, 2, 3])))
def test_which_multiple_none(self):
"""tests the which_multiple() function"""
self.assertEquals(0, len(util.which_multiple([])))
