-
Notifications
You must be signed in to change notification settings - Fork 0
/
classify.py
60 lines (42 loc) · 1.56 KB
/
classify.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
from collections import defaultdict
import sys
from mrjob.protocol import JSONProtocol
from numpy import product
def classify(category_probs, cond_probs, features):
posteriors = defaultdict(float)
for category in category_probs:
posteriors[category] = compute_posterior(category, category_probs[category], cond_probs, features)
return max(posteriors, key=posteriors.get)
#not the actual posterior, but proportional to it independent of the category
def compute_posterior(category, category_prob, cond_probs, features):
return category_prob * product([cond_probs[(category, feature)] for feature in features])
def parse_in_file(name):
source = open(name, 'r')
parsed = [parse_in_line(line)for line in source]
source.close()
return parsed
def parse_in_line(line):
return line.split()
def parse_prob_file(name):
cond_probs = defaultdict(float)
category_probs = defaultdict(float)
source = open(name, 'r')
for line in source:
pair, prob = read(line)
category, feature = pair
if feature is None:
category_probs[category] = prob
else:
cond_probs[pair] = prob
source.close()
return category_probs, cond_probs
def read(string):
list_, prob = JSONProtocol.read(string)
return tuple(list_), prob
if __name__ == "__main__":
prob_file = sys.argv[1]
in_file = sys.argv[2]
category_probs, cond_probs = parse_prob_file(prob_file)
data = parse_in_file(in_file)
for features in data:
print(classify(category_probs, cond_probs, features))