def create_base(path, host, port, db_name):
base = TextBase(host, port, db_name)
for theme in os.listdir(path):
docs_dir = os.path.join(path, theme)
for doc in os.listdir(docs_dir):
remove_transfer(os.path.join(docs_dir, doc))
base.append(doc, formalize(os.path.join(docs_dir, doc)))
base._normalize()
return base
def check_svm(self):
host = self.pargs.host or MONGODB_BACKEND_SETTINGS['host']
port = self.pargs.port or MONGODB_BACKEND_SETTINGS['port']
name = self.pargs.name or MONGODB_BACKEND_SETTINGS['database']
from mysvm import SVM
base = TextBase(host, port, name)
input, target = base.to_lists()
s = SVM(input, target)
#testing
from pymongo import Connection
connection = Connection(host=host, port=port)
db = connection[name]
collection = db['test_collection2']
docs = collection.find()
count = 0.0
for doc in docs:
res = s.predict(base.to_list(doc))
if doc['_class_name'] == base.classes[int(res)]:
count += 1
print count / docs.count() * 100
