In [2]:

    

from _collections import defaultdict
import time
import timeit

from numpy.linalg import norm
import scipy.optimize

import numpy as np


def parseData(fname):
  for l in open(fname):
    yield eval(l)
    
def parseTxt(fname):
  for l in open(fname):
    yield l.strip().split(" ")

print "Reading train..."
train = list(parseData("homework6_7/train.json"))
print "done"

allXs = []
allYs = []
for l in train:
  user, item, rating = l['reviewerID'], l['itemID'], l['rating']
  allXs.append([user, item])
  allYs.append(float(rating))

print "Reading test..."
test = np.array(list(parseTxt("homework6_7/labeled_Rating.txt")))
print "done"


def miniFunc(Data, Alpha, BetaU, BetaI, Lambd):
    part1 = 0
    for [u, i], Rui in Data:
        part1 += ((Alpha + BetaU[u] + BetaI[i] - Rui) ** 2)
    
    part2 = 0
    for u in BetaU:
        part2 += (BetaU[u] ** 2)
    for i in BetaI:
        part2 += (BetaI[i] ** 2)
    
    return part1 + Lambd * part2   


oldVal = 0
lambd = 1
alpha = 0
X = allXs
y = allYs
betaU = defaultdict(int)
betaI = defaultdict(int)
Ntrain = len(y)

data = zip(X, y)

while True:
    lastAlpha = alpha
    lastBetaU = betaU
    lastBetaI = betaI
    
    # Alpha stage
    alpha = 0
    for [u, i], Rui in data:
        bu = betaU[u]
        bi = betaI[i]
        alpha += Rui - (bu + bi)
    alpha = alpha / Ntrain
    
    # BetaU stage 
    Iu = defaultdict(int)
    betaU = defaultdict(int)
    for [u, i], Rui in data:
        betaU[u] += (Rui - (alpha + betaI[i]))
        Iu[u] += 1
    for u in betaU:
        betaU[u] = betaU[u] / (lambd + Iu[u])

    # BetaI stage 
    Ii = defaultdict(int)
    betaI = defaultdict(int)
    for [u, i], Rui in data:
        betaI[i] += (Rui - (alpha + betaU[u]))
        Ii[i] += 1
    for i in betaI:
        betaI[i] = betaI[i] / (lambd + Ii[i])
        
    newVal = miniFunc(data, alpha, betaU, betaI, lambd)
    
    if lastAlpha != 0 and oldVal < newVal:
        alpha = lastAlpha
        betaU = lastBetaU
        betaI = lastBetaI
        break
        
    oldVal = newVal
    
print alpha


    

    




Reading train...
done
Reading test...
done
3.5194082852

In [3]:

    

item = 'I102776733'
user = 'U566105319'


    

In [4]:

    

print "Alpha: ", alpha
print "BetaI of ", item, ": ", betaI[item]
print "BetaU of ", user, ": ", betaU[user]


    

    




Alpha:  3.5194082852
BetaI of  I102776733 :  0.171752035867
BetaU of  U566105319 :  -1.03076735296

In [5]:

    

mse = 0
for u, i, Rui in test:
    p = alpha + betaU[u] + betaI[i]
    mse += (p - float(Rui)) ** 2
mse = mse / len(test)

print "Mse: ", mse


    

    




Mse:  2.19889993422

In [45]:

    




    

In [6]:

    

user1 = 'U229891973'
user2 = 'U622491081'

A = set([i for (u, i) in allXs if u == user1])
B = set([i for (u, i) in allXs if u == user2])


    

In [7]:

    

print len(A), len(B)


    

    




4 3

In [8]:

    

Jaccard1 = len(A.intersection(B)) * 1.0 / len((A.union(B)))
print Jaccard1


    

    




0.75

In [9]:

    

user3 = 'U622491081'

dictU = defaultdict(list)
for u, i in allXs:
    dictU[u].append(i)


    

In [10]:

    

A = set(dictU[user3])

bestJac = 0
bestU = []

for u in dictU:
    if u == user3:
        continue
        
    B = set(dictU[u])
    jacc = len(A.intersection(B)) * 1.0 / len(A.union(B))
    
    if jacc > bestJac:
        bestU = [u]
        bestJac = jacc
    elif jacc == bestJac:
        bestU.append(u)
        
print bestU, bestJac


    

    




['U359587607', 'U687939146', 'U096951499', 'U296575297', 'U387971231', 'U300899166'] 1.0

In [10]:

    




    

In [11]:

    

user4 = 'U639726733'

dictI = defaultdict(list)
for u, i in allXs:
    dictI[i].append(u)


    

In [12]:

    

items = dictU[user4]

for item in items:
    A = set(dictI[item])

    bestJac = 0
    bestI = []

    for i in dictI:
        if i in items:
            continue
            
        B = set(dictI[i])
        jacc = len(A.intersection(B)) * 1.0 / len(A.union(B))

        if jacc > bestJac:
            bestI = [i]
            bestJac = jacc
        elif jacc == bestJac:
            #bestI += [i]
            bestI = [i]

    print item, bestI, bestJac
    print "-------------"


    

    




I827118969 ['I368057136'] 0
-------------
I988644602 ['I368057136'] 0
-------------
I958777870 ['I368057136'] 0
-------------
I616454620 ['I368057136'] 0
-------------
I970119134 ['I970165134'] 0.2
-------------

In [12]:

    




    

In [ ]: