def getLabel(self, value):
feature = FeatureFactory()
feature.createFeature(value, "")
dict = {};
dict['attributes'] = {}
attributes = []
line = feature.datatable[0]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
res = self.model.predict(dict)
r = max(res.iterkeys(),key=lambda k:res[k])
return r
def getLabel(self, value):
feature = FeatureFactory()
feature.createFeature(value, "")
dict = {}
dict['attributes'] = {}
attributes = []
line = feature.datatable[0]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
res = self.model.predict(dict)
r = max(res.iterkeys(), key=lambda k: res[k])
return r
def getClass(setting, value):
setting = setting.decode("string-escape")
#print setting
classifier = pickle.loads(setting)
feature = FeatureFactory()
feature.createFeature(value, "")
dict = {}
dict['attributes'] = {}
attributes = []
line = feature.datatable[0]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
res = classifier.predict(dict)
r = max(res.iterkeys(), key=lambda k: res[k])
return r
def getClass(setting, value):
setting = setting.decode("string-escape")
#print setting
classifier = pickle.loads(setting)
feature = FeatureFactory()
feature.createFeature(value, "")
dict = {};
dict['attributes'] = {}
attributes = []
line = feature.datatable[0]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
res = classifier.predict(dict)
r = max(res.iterkeys(), key=lambda k: res[k])
return r
class IDCTClassifier(PartitionClassifierType):
def __init__(self):
self.path = "./"
self.featureFactory = FeatureFactory()
def addTrainingData(self, value, label):
self.featureFactory.createFeature(value, label)
def learnClassifer(self):
model = NaiveBayes()
dict = {}
dict['cases'] = 1
attributes = []
for j in range(len(self.featureFactory.datatable)):
dict = {}
dict['cases'] = 1
dict['attributes'] = {}
line = self.featureFactory.datatable[j]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
dict['label'] = self.featureFactory.classes[j]
model.add_instances(dict)
model.set_real(attributes)
model.train()
self.model = model
return pickle.dumps(model).encode('string_escape')
def getLabel(self, value):
feature = FeatureFactory()
feature.createFeature(value, "")
dict = {}
dict['attributes'] = {}
attributes = []
line = feature.datatable[0]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
res = self.model.predict(dict)
r = max(res.iterkeys(), key=lambda k: res[k])
return r
class IDCTClassifier(PartitionClassifierType):
def __init__(self):
self.path = "./"
print "building classifier"
self.featureFactory = FeatureFactory()
def addTrainingData(self, value, label):
self.featureFactory.createFeature(value, label)
def learnClassifer(self):
model = NaiveBayes()
dict = {};
dict['cases'] = 1
attributes = []
for j in range(len(self.featureFactory.datatable)):
dict = {};
dict['cases'] = 1
dict['attributes'] = {}
line = self.featureFactory.datatable[j]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
dict['label'] = self.featureFactory.classes[j]
model.add_instances(dict)
model.set_real(attributes)
model.train()
self.model = model
return pickle.dumps(model).encode('string_escape')
def getLabel(self, value):
feature = FeatureFactory()
feature.createFeature(value, "")
dict = {};
dict['attributes'] = {}
attributes = []
line = feature.datatable[0]
for i in range(len(line)):
dict['attributes'][str(i)] = line[i]
attributes.append(str(i))
res = self.model.predict(dict)
r = max(res.iterkeys(),key=lambda k:res[k])
return r
def main(argv):
# Set default values
similarityMeasure = None
predictionModel = None
numTrials = 0
buildClean = False
# Check for 'help' argument
if argv[1] == 'help':
printArgsHelp()
sys.exit(0)
# Check for valid arguments
if len(argv) < 4:
sys.stderr.write('Too few arguments! Try running with the \'help\' argument.\n')
sys.exit(1)
if argv[1] not in similarityMeasureStrings:
sys.err.write('Invalid similarity measure! Try running with the \'help\' argument.\n')
sys.exit(1)
if argv[2] not in predicitonModelStrings:
sys.err.write('Invalid prediction model! Try running with the \'help\' argument.\n')
sys.exit(1)
try:
numTrials = int(argv[3])
except ValueError:
sys.err.write('Please provide int value for \'trials\' argument. Try running with the \'help\' argument.\n')
sys.exit(1)
if int(argv[3]) < 0:
sys.err.write('Invalid integer for \'trials\'! Try running with the \'help\' argument.\n')
sys.exit(1)
if len(argv) > 4 and argv[4] != 'clean':
sys.err.write('Invalid value for buildClean! Try running with the \'help\' argument.\n')
sys.exit(1)
if len(argv) > 4 and argv[4] == 'clean':
buildClean = True
# Generate Yelp data either
# by parsing Jsons or loading from .bins
# yelpData currently contains:
# a map from user -> friends
# a map from business -> users who rated that business
# TODO: expand what yelp data contains as necessary for other sim measures
if buildClean:
yelpData = yelp_json_parser.parseJsons(businessJson='NV_business.json', reviewJson='NV_review.json', userJson='NV_user.json')
else:
yelpData = yelp_json_parser.loadFromFile()
friendshipMap = yelpData[0]
businessReviews = yelpData[1]
if buildClean:
yelpGraph = yelpData[5]
degreeCentrality = yelpData[2]
closenessCentrality = yelpData[3]
betweennessCentrality = yelpData[4]
print "Betweenness Centralities"
print len(degreeCentrality)
# Create appropriate similarity measure (with necessary yelp data) and
# either calculate similarities from scratch (buildClean == True) or
# load similarities from file (buildClean == False)
similarityScores = dict()
if argv[1] == 'foverlap':
similarityMeasure = FriendshipOverlapSimilarity(friendshipMap)
elif argv[1] == 'community':
similarityMeasure = CommunitySimilarity(yelpGraph)
elif argv[1] == 'commute':
similarityMeasure = CommuteTimeSimilarity()
elif argv[1] == 'pagerank':
similarityMeasure = PageRankSimilarity(friendshipMap)
elif argv[1] == 'featureDist':
factory = FeatureFactory((degreeCentrality, closenessCentrality, betweennessCentrality))
vectors = factory.getFeatureMatrix()
similarityMeasure = FeatureDistanceSimilarity(vectors)
elif argv[1] == 'all':
similarityMeasure = FriendshipOverlapSimilarity(friendshipMap)
similarityMeasure.calculateSimilarities()
similarityScores = similarityMeasure.similarities
predictionsFOverlap = RegressorUtil.runRegressor(similarityScores, businessReviews, KNNRegressor())
similarityMeasure = CommunitySimilarity(yelpGraph)
similarityMeasure.calculateSimilarities()
similarityScores = similarityMeasure.similarities
predictionsCommunity = RegressorUtil.runRegressor(similarityScores, businessReviews, KNNRegressor())
similarityMeasure = PageRankSimilarity(friendshipMap)
similarityMeasure.calculateSimilarities()
similarityScores = similarityMeasure.similarities
predictionsPageRank = RegressorUtil.runRegressor(similarityScores, businessReviews, KNNRegressor())
factory = FeatureFactory((degreeCentrality, closenessCentrality, betweennessCentrality))
vectors = factory.getFeatureMatrix()
similarityMeasure = FeatureDistanceSimilarity(vectors)
similarityMeasure.calculateSimilarities()
similarityScores = similarityMeasure.similarities
predictionsFeatureDist = RegressorUtil.runRegressor(similarityScores, businessReviews, KNNRegressor())
predictions = RegressorUtil.averagePredictions((predictionsFOverlap,predictionsCommunity,predictionsPageRank,predictionsFeatureDist))
RegressorUtil.evaluateRegressor(predictions, 'All', 'All')
sys.exit(0)
if buildClean:
similarityMeasure.calculateSimilarities()
else:
similarityMeasure.loadFromFile()
similarityScores = similarityMeasure.similarities
print len(similarityScores)
# Create appropriate prediction model and
# generate list of predictions
if argv[2] == 'baseline':
predictionModel = RandomRegressor(1,5)
elif argv[2] == 'knn':
predictionModel = KNNRegressor()
# Once similarities are calculated, the true ratings are parsed,
# and the prediction model is chosen, we then run our regression model
# to generate our predictions for each business-user pair
predictions = RegressorUtil.runRegressor(similarityScores, businessReviews, predictionModel)
# Once all the predictions have been calculated, we evaluate the accuracy of
# our system and report error statistics
RegressorUtil.evaluateRegressor(predictions, predictionModel.nameLabel, similarityMeasure.nameLabel)
def __init__(self):
self.path = "./"
self.featureFactory = FeatureFactory()
def __init__(self):
self.path = "./"
self.featureFactory = FeatureFactory()
def __init__(self):
self.path = "./"
print "building classifier"
self.featureFactory = FeatureFactory()
