def getFullInfluenceScore():
partialScoreCollection = getClusterInfluencerScoreCollection()
userClusterContributionCollection = getUserClusterContributionCollection()
fbfriends = getFriendsCollection()
fullScoreCollection = getFullInfluenceScoreCollection()
for friend in fbfriends.find():
fid = friend['id']
partial = partialScoreCollection.find_one({'_id': fid})
contribution = userClusterContributionCollection.find_one({'_id': fid})
score = []
for i in range(len(contribution['contribution'])):
avg = 1.0
try:
avg *= sum(partial['cluster'][str(i)]) / len(
partial['cluster'][str(i)])
except ZeroDivisionError:
avg = 0.0
score.append(avg * contribution['contribution'][i])
if len(score) > 0:
document = {}
document['_id'] = fid
document['name'] = friend['name']
document['score'] = score
fullScoreCollection.insert(document)
print friend['name'], score
def getFullInfluenceScore():
partialScoreCollection = getClusterInfluencerScoreCollection()
userClusterContributionCollection = getUserClusterContributionCollection()
fbfriends = getFriendsCollection()
fullScoreCollection = getFullInfluenceScoreCollection()
for friend in fbfriends.find():
fid = friend['id']
partial = partialScoreCollection.find_one({'_id' : fid})
contribution = userClusterContributionCollection.find_one({'_id' : fid})
score = []
for i in range(len(contribution['contribution'])):
avg = 1.0
try:
avg *= sum(partial['cluster'][str(i)])/len(partial['cluster'][str(i)])
except ZeroDivisionError:
avg = 0.0
score.append(avg*contribution['contribution'][i])
if len(score) > 0:
document = {}
document['_id'] = fid
document['name'] = friend['name']
document['score'] = score
fullScoreCollection.insert(document)
print friend['name'], score
def getClusterContribution():
clusterInfluencerScoreCollection = getClusterInfluencerScoreCollection()
userClusterContributionCollection = getUserClusterContributionCollection()
userClusterContributionCollection.drop()
for scores in clusterInfluencerScoreCollection.find():
document = {}
document['_id'] = scores['_id']
document['contribution'] = []
for cscore in sorted(scores['cluster'].keys()):
document['contribution'].append(len(scores['cluster'][cscore]))
document['contribution'] = [i*1.0/sum(document['contribution']) for i in document['contribution'] if sum(document['contribution']) > 0]
print 'Calculated for', userClusterContributionCollection.insert(document)
def getPartialInfluenceScore():
clusterInfoCollection = getPagesClusterInfoCollection()
clusterInfluencerCollection = getClusterInfluencerScoreCollection()
friendsCollection = getFriendsCollection()
clusterInfluencerCollection.drop()
clusterNumber = len(clusterInfoCollection.distinct('cluster'))
for friend in friendsCollection.find():
_id = friend['id']
document = {}
document['_id'] = _id
document['cluster'] = {}
for i in range(clusterNumber):
document['cluster'][str(i)] = []
clusterInfluencerCollection.insert(document)
pagesCursor = clusterInfoCollection.find({"count": {"$gt": 3}})
epoch = datetime.datetime.utcfromtimestamp(0)
dt = 7 * 24 * 60 * 60
scores = []
done = 0
for page in pagesCursor:
users = page['people']
try:
users.sort(key=lambda x: x['created_time'])
cluster = page['cluster']
liketime = []
for user in users:
liketime.append((user['created_time'] - epoch).total_seconds())
back = 0
done += 1
print done
for user in users:
userId = user['id']
timeahead = (user['created_time'] - epoch).total_seconds() + dt
timeback = (user['created_time'] - epoch).total_seconds() - dt
ahead = bisect.bisect_right(liketime, timeahead)
score = ahead - back
back += 1
# print userId, cluster, score
clusterInfluencerCollection.update(
{'_id': userId},
{'$push': {
'cluster.' + str(cluster): score
}},
upsert=False)
except:
print "hmmm"
def getPartialInfluenceScore():
clusterInfoCollection = getPagesClusterInfoCollection()
clusterInfluencerCollection = getClusterInfluencerScoreCollection()
friendsCollection = getFriendsCollection()
clusterInfluencerCollection.drop()
clusterNumber = len(clusterInfoCollection.distinct('cluster'))
for friend in friendsCollection.find():
_id = friend['id']
document = {}
document['_id'] = _id
document['cluster'] = {}
for i in range(clusterNumber):
document['cluster'][str(i)] = []
clusterInfluencerCollection.insert(document)
pagesCursor = clusterInfoCollection.find({"count": {"$gt": 3}})
epoch = datetime.datetime.utcfromtimestamp(0)
dt = 7 * 24 * 60 * 60
scores = []
done = 0
for page in pagesCursor:
users = page['people']
try:
users.sort(key=lambda x: x['created_time'])
cluster = page['cluster']
liketime = []
for user in users:
liketime.append((user['created_time'] - epoch).total_seconds())
back = 0
done += 1
print done
for user in users:
userId = user['id']
timeahead = (user['created_time'] - epoch).total_seconds() + dt
timeback = (user['created_time'] - epoch).total_seconds() - dt
ahead = bisect.bisect_right(liketime, timeahead)
score = ahead - back
back += 1
# print userId, cluster, score
clusterInfluencerCollection.update({'_id': userId}, {'$push': {'cluster.' + str(cluster): score}},
upsert=False)
except:
print "hmmm"
def getClusterContribution():
clusterInfluencerScoreCollection = getClusterInfluencerScoreCollection()
userClusterContributionCollection = getUserClusterContributionCollection()
userClusterContributionCollection.drop()
for scores in clusterInfluencerScoreCollection.find():
document = {}
document['_id'] = scores['_id']
document['contribution'] = []
for cscore in sorted(scores['cluster'].keys()):
document['contribution'].append(len(scores['cluster'][cscore]))
document['contribution'] = [
i * 1.0 / sum(document['contribution'])
for i in document['contribution']
if sum(document['contribution']) > 0
]
print 'Calculated for', userClusterContributionCollection.insert(
document)