def classify(self,classify_file): content=Preprocess([]).singlefile(open(classify_file)) result=0 max_probability=-2**32 for cur_class in range(self.class_number): cur_probability=self.p_classes[cur_class] for word in content.split(): if word in self.map: cur_probability+=self.map[word][cur_class] else: cur_probability+=self.p_unknown[cur_class] if cur_probability>max_probability: max_probability=cur_probability result=cur_class return "spam" if result==0 else "ham"
import sys
import os
from nbpreprocess import Preprocess
from nbclassify import Nbclassify
from nbtraing import Nbtraing
if __name__=="__main__":
##read a direcotry and use nb classify and evaluate the result using f-measure
p=Preprocess(sys.argv[1:3])
p.process()
nbtraing=Nbtraing(sys.argv[1:3])
nbtraing.train()
classifier=Nbclassify("nb.model");
test_path=sys.argv[3]
tp=0;fp=0;fn=0;
for f in os.listdir(test_path):
result=classifier.classify(os.path.join(test_path,f))
origin="spam" if "spm" in f else "ham"
if origin==result=="spam": tp+=1
if origin=="ham" and result=="spam":fp+=1
if origin=="spam" and result=="ham":fn+=1
p=float(tp)/(tp+fp)
r=float(tp)/(tp+fn)
print 2*p*r/(r+p)