forked from nico/collectiveintelligence-book
-
Notifications
You must be signed in to change notification settings - Fork 1
/
treepredict.py
200 lines (162 loc) · 5.52 KB
/
treepredict.py
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
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
import collections
class decisionnode(object):
def __init__(self, col=-1, value=None, results=None, tb=None, fb=None):
self.col = col # colum index of value to test
self.value = value # reference value
self.results = results # stores results in leafs, empty for inner nodes
self.tb = tb # child on true branch
self.fb = fb # child on false branch
def divideset(rows, column, value):
split_function = None
if isinstance(value, int) or isinstance(value, float):
split_function = lambda row: row[column] >= value
else:
split_function = lambda row: row[column] == value
# There has to be a `partition` or `group` function somewhere
set1 = [row for row in rows if split_function(row)]
set2 = [row for row in rows if not split_function(row)]
return (set1, set2)
def uniquecounts(rows):
results = collections.defaultdict(int)
for row in rows:
# Result is last column
r = row[len(row) - 1]
results[r] += 1
return dict(results)
def giniimpurity(rows):
"""Returns probability that a randomly placed item will end up in the wrong
category. A low result means that stuff is categorized well."""
total = len(rows)
counts = uniquecounts(rows)
imp = 0
# O(n^2) in number of categories
for k1 in counts:
p1 = float(counts[k1])/total # XXX: These loops can be written more nicely
for k2 in counts:
if k1 == k2: continue
p2 = float(counts[k2])/total
imp += p1*p2
return imp
def entropy(rows):
from math import log
log2 = lambda x: log(x)/log(2)
results = uniquecounts(rows)
ent = 0.0
for r in results:
p = float(results[r])/len(rows)
ent -= p*log2(p)
return ent
def variance(rows):
if len(rows) == 0: return 0
data = [float(row[len(row) - 1]) for row in rows]
mean = sum(data) / len(data)
# this gives indexoutofbounds in zillow example
#variance = sum([(d-mean)**2 for d in data]) / (len(data) - 1)
variance = sum([(d-mean)**2 for d in data]) / len(data)
return variance
def buildtree(rows, scorefun=entropy):
if len(rows) == 0: return decisionnode()
current_score = scorefun(rows)
best_gain = 0.0
best_criteria = None
best_sets = None
column_count = len(rows[0]) - 1 # last column is result
for col in range(0, column_count):
# find different values in this column
column_values = set([row[col] for row in rows])
# for each possible value, try to divide on that value
for value in column_values:
set1, set2 = divideset(rows, col, value)
# Information gain
p = float(len(set1)) / len(rows)
gain = current_score - p*scorefun(set1) - (1-p)*scorefun(set2)
if gain > best_gain and len(set1) > 0 and len(set2) > 0:
best_gain = gain
best_criteria = (col, value)
best_sets = (set1, set2)
if best_gain > 0:
trueBranch = buildtree(best_sets[0])
falseBranch = buildtree(best_sets[1])
return decisionnode(col=best_criteria[0], value=best_criteria[1],
tb=trueBranch, fb=falseBranch)
else:
return decisionnode(results=uniquecounts(rows))
def printtree(tree, indent=''):
if tree.results != None: # leaf node
print tree.results
else:
print '%s:%s?' % (tree.col, tree.value)
print indent + 'T->',
printtree(tree.tb, indent + ' ')
print indent + 'F->',
printtree(tree.fb, indent + ' ')
def classify(observation, tree):
if tree.results != None: # leaf
return tree.results
else:
v = observation[tree.col]
branch = None
if isinstance(v, int) or isinstance(v, float):
if v >= tree.value: branch = tree.tb
else: branch = tree.fb
else:
if v == tree.value: branch = tree.tb
else: branch = tree.fb
return classify(observation, branch)
def prune(tree, mingain):
# recurse
if tree.tb.results == None: prune(tree.tb, mingain)
if tree.fb.results == None: prune(tree.fb, mingain)
# merge leaves (potentionally)
if tree.tb.results != None and tree.fb.results != None:
tb, fb = [], []
for v, c in tree.tb.results.iteritems(): tb += [[v]] * c
for v, c in tree.fb.results.iteritems(): fb += [[v]] * c
p = float(len(tb)) / len(tb + fb)
delta = entropy(tb+fb) - p*entropy(tb) - (1-p)*entropy(fb)
if delta < mingain:
tree.tb, tree.fb = None, None
tree.results = uniquecounts(tb + fb)
# 'missing data classify'
def mdclassify(observation, tree):
if tree.results != None: # leaf
return tree.results
else:
v = observation[tree.col]
if v == None:
tr = mdclassify(observation, tree.tb)
fr = mdclassify(observation, tree.fb)
tcount = sum(tr.values())
fcount = sum(fr.values())
tw = float(tcount)/(tcount + fcount)
fw = float(fcount)/(tcount + fcount)
result = collections.defaultdict(int)
for k, v in tr.iteritems(): result[k] += v*tw
for k, v in fr.iteritems(): result[k] += v*fw
return dict(result)
else:
branch = None
if isinstance(v, int) or isinstance(v, float):
if v >= tree.value: branch = tree.tb
else: branch = tree.fb
else:
if v == tree.value: branch = tree.tb
else: branch = tree.fb
return classify(observation, branch)
def testdata():
def cleanup(s):
s = s.strip()
try:
return int(s)
except ValueError:
return s
return [map(cleanup, line.split('\t'))
for line in open('decision_tree_example.txt')]
if __name__ == '__main__':
tree = buildtree(testdata())
printtree(tree)
print classify(['(direct)', 'USA', 'yes', 5], tree)
prune(tree, 0.1)
printtree(tree)
prune(tree, 1.0)
printtree(tree)