def testReview(self, reviewJSON): review = f.getJson(reviewJSON) userID = f.getUser(review) businessID = f.getBusiness(review) rate = f.getRate(review) prediction = self.predictKNN(userID, businessID) return (userID, rate == prediction, rate, prediction)
def __init__(self, numNeighbors, reviewFilePath, featureType):
self.numNeighbors = numNeighbors
self.featureType = featureType
self.busiID2Busi = {}
self.userID2User = {}
self.userCount=0
self.businessCount=0
reviews = open(reviewFilePath, 'r')
for r in reviews:
reviewJson = f.getJson(r)
userID = f.getUser(reviewJson)
businessID = f.getBusiness(reviewJson)
text = f.getText(reviewJson)
rate = f.getRate(reviewJson)
if businessID not in self.busiID2Busi.keys():
busiObj = Business(businessID, numNeighbors)
self.busiID2Busi[businessID]=busiObj
self.businessCount+=1
else:
busiObj = self.busiID2Busi[businessID]
if userID not in self.userID2User.keys():
userObj = User(userID)
self.userCount+=1
self.userID2User[userID]=userObj
else:
userObj = self.userID2User[userID]
userObj.addReview(text)
busiObj.addUser(userObj,rate)
reviews.close()
for b in self.busiID2Busi.values():
b.trainKNN(self.featureType)