1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
|
# Copyright (c) 2004-2007 Bartek Wilczynski and Norbert Dojer; All Rights Reserved.
#
# This software is distributable under the terms of the GNU
# General Public License (GPL) v2, the text of which can be found at
# http://www.gnu.org/copyleft/gpl-2.0.txt. Installing, importing or otherwise
# using this module constitutes acceptance of the terms of this License.
#
# Disclaimer
#
# This software is provided "as-is". There are no expressed or implied
# warranties of any kind, including, but not limited to, the warranties
# of merchantability and fittness for a given application. In no event
# shall the authors be liable for any direct, indirect, incidental,
# special, exemplary or consequential damages (including, but not limited
# to, loss of use, data or profits, or business interruption) however
# caused and on any theory of liability, whether in contract, strict
# liability or tort (including negligence or otherwise) arising in any way
# out of the use of this software, even if advised of the possibility of
# such damage.
#
import math
from score import score
class MDL(score):
def graph_score(self,number_of_potential_parents,gene_vertex,weights_of_parents,number_of_data_points): # g(Pa)
graph_structure=(1+len(weights_of_parents))*math.log(number_of_potential_parents+1,2) # = |Pa| * log n
graph_cpd=0.5*(max(1.1,gene_vertex.n_disc)-1)*math.log(number_of_data_points+1,2) # = 1/2 * log N * (k_x - 1)
if gene_vertex.cpd in ['or','and','preproc']:
graph_cpd*=1+sum(weights_of_parents)/2.0
else:
for weight in weights_of_parents:
graph_cpd*=weight
return graph_structure+graph_cpd
def lower_bound_for_data_score(self,selected_data):
if selected_data.vertex.n_disc:
return 0.0
score=0.0
for d in selected_data.data:
try:
score -= d[0]*math.log(d[0],2) + (1-d[0])*math.log(1-d[0],2)
except:
pass
return score*self.data_factor
def data_score(self,selected_data):
stats_all,stats_parents = selected_data.stats()
score = 0.0
if selected_data.vertex.cpd in ['or','and']: # continuous
prod_in=(selected_data.vertex.cpd=='or')
prod_out=1-prod_in
probabilities=self.cpd_andor(len(selected_data.parents),stats_all,stats_parents,prod_in)
for a in stats_parents.keys():
prob_a=probabilities[-1] # last element
for i,av in enumerate(a):
if av==prod_in:
prob_a*=probabilities[i]
try:
score-=stats_all[a+(prod_out,)]*math.log(prob_a,2)
except:
pass
try:
score-=stats_all[a+(prod_in,)]*math.log(1-prob_a,2)
except:
pass
else:
for a,cv in stats_all.items():
numbers_of_parents = a[:-1] # except the last one
cp = stats_parents[numbers_of_parents]
try:
sc = math.log(1.0*cp/cv,2) # - log(cv/cp)
score+=cv*sc
except ZeroDivisionError:
pass
return score*self.data_factor
|