recommendation.py

from scipy.stats.stats import pearsonr
import numpy as np
import pymysql.cursors
import datetime
import traceback
import json
import operator
import ConfigParser
import random
from search import searchQuery

config = ConfigParser.ConfigParser()
config.read('db.cfg')

connection = pymysql.connect(host=config.get('database','host'),
							 user=config.get('database','username'),
							 password=config.get('database','password'),
							 db = config.get('database','db'),
							 charset = 'utf8mb4',
							 cursorclass=pymysql.cursors.DictCursor)

ratingsData = {}

def getRatings(user):
	try:
		with connection.cursor() as cursor:
			ratingsData[user] = {}
			sql = """select `TopicId`, `Weight` from `UserInterests` 
					where `UserId` = %s"""
			rowCount = cursor.execute(sql, user)
			if rowCount > 0:
				topics = cursor.fetchall()
				for row in topics:
					sqlTopic = """select `Name` from `Topics` where `Id` = %s"""
					rowCountTopic = cursor.execute(sqlTopic, row[u'TopicId'])
					if rowCountTopic > 0:
						result = cursor.fetchall()[0]
						topic = str(result[u'Name'])
						weight = row[u'Weight']
						ratingsData[user][topic] = weight
			return ratingsData[user]

	except Exception, e:
		print traceback.print_exc()
		return -1

def pearson_correlation(user1, user2):
	print "pearson_correlation"
	user1RatingsDict = getRatings(user1)
	user2RatingsDict = getRatings(user2)
	common_topics = [topic for topic in user1RatingsDict if user1RatingsDict[topic] != 0 and user2RatingsDict[topic] != 0]
	user1Ratings = [user1RatingsDict[topic] for topic in common_topics]
	user2Ratings = [user2RatingsDict[topic] for topic in common_topics]
	return pearsonr(user1Ratings, user2Ratings)[0]

def getAllUsers():
	allUsers = []
	print "In allUsers"
	try:
		with connection.cursor() as cursor:
			sql = """select `Id` from `Users` LIMIT 1000"""
			rowCount = cursor.execute(sql)
			if rowCount > 0:
				result = cursor.fetchall()
			for row in result:
				allUsers.append(row[u'Id'])
		#print allUsers
		allUsers = random.sample(allUsers, 10)
		return allUsers

	except Exception, e:
		print traceback.print_exc()
		return -1

def getSimilarUsers(user):
	try:
		print "In getSimilarUsers"
		user = int(user)
		allUsers = getAllUsers()
		scores = [(pearson_correlation(user, otherUser), otherUser)
		 			for otherUser in allUsers if otherUser != user]

		scores.sort()
		scores.reverse()
		similarUsers = [row[1] for row in scores]
		return similarUsers[0:5]
	except Exception, e:
		print traceback.print_exc()
		return -1

def getSimilarUsersForUI(user):
	try:
		print "In getSimilarUsers"
		user = int(user)
		allUsers = getAllUsers()
		scores = [(pearson_correlation(user, otherUser), otherUser)
		 			for otherUser in allUsers if otherUser != user]

		scores.sort()
		scores.reverse()
		scores = scores[0:5]		
		similarUsers = [row[1] for row in scores]
		
		users = []
		for user in similarUsers:
			with connection.cursor() as cursor:
				sql = """select `Id`, `DisplayName` from `Users` where Id = %s"""
				rowCount = cursor.execute(sql, (user))
				res = cursor.fetchone()
				users.append(res)
		
		for userr in users:
			userr[u'DisplayName'] = userr[u'DisplayName'].title()
			userr[u'CurrentUserId'] = user 
		
		return users
	except Exception, e:
		print traceback.print_exc()
		return -1


def getTopicsToRecommend(user):
	topics = []
	interestedTopics = []
	similarUsers = getSimilarUsers(user)
	#userIDs = [record[1] for record in similarUsers]
	for userID in similarUsers:
		topics.extend(ratingsData[userID].items())

	topics = sorted(topics, key=operator.itemgetter(1))[::-1]

	interestedTopics.extend(row for row in topics 
						if row[0] not in [item[0] for item in interestedTopics])

	return interestedTopics[0:3]

def recommendQuestions(user):
	try:
		user = int(user)
		topics = getTopicsToRecommend(user)
		print topics
		questions = []
		for topic in topics:
			questions.extend(searchQuery(topic[0], 2))
		#print questions
		return questions
	except Exception, e:
		print traceback.print_exc()
		return -1



if __name__ == "__main__":
	q =  recommendQuestions('821742')
	for i in q:
			print i[u'Id']
	print getSimilarUsers(int('821742'))