-
Notifications
You must be signed in to change notification settings - Fork 0
/
USER_MOVIE_NETWORK_gaussian.py
333 lines (254 loc) · 9.93 KB
/
USER_MOVIE_NETWORK_gaussian.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
# -*- coding: utf-8 -*-
#Author: Greg Murray
#Title: Gaussian User-Movie Network Recommendation Engine
from pyspark.storagelevel import StorageLevel
import math
sc = spark.sparkContext
#small
#rdd=sc.textFile('/Users/gregmurray/Documents/BigData/movie_rec_engine/Final_Package/data_sources/ratings_sample_tiny.csv').persist(storageLevel=StorageLevel.MEMORY_AND_DISK)
#99k rows
rdd=sc.textFile('/Users/gregmurray/Documents/BigData/movie_rec_engine/Final_Package/Testing/ml-20m/ratings_train_99k.csv').persist(storageLevel=StorageLevel.MEMORY_AND_DISK)
#test2
#rdd=sc.textFile('/Users/gregmurray/Documents/BigData/movie_rec_engine/Final_Package/data_sources/ratings_sample_test2.csv').persist(storageLevel=StorageLevel.MEMORY_AND_DISK)
#______________________________________________________________________________#
#Functions
def li(v): return [v]
def app(a,b):
a.append(b)
return a
def ext(a,b):
a.extend(b)
return a
#standard deviation of each user
import statistics as st
def sd1(v):
if len(v[1])==1:
return (v, 0)
else:
return (v, round(st.stdev(v[1]),3))
#mean of each movie or user.
def mean(v):
if len(v[0][1])==1:
return ((v[0][0], (v[0][1][0], v[1])), v[0][1])
else:
return (( v[0][0], (round(st.mean(v[0][1]),3), v[1])), v[0][1] )
def normalize(v):
norm_ratings = []
mu = v[0][1][0]
sigma = v[1][1][1]
for r in v[1]:
z = (r-mu)/sigma
norm_ratings.append(z)
return (v[0], norm_ratings)
def get_wrdv_orig(arr):
res=[]
for tup in arr:
wrdv = round((1+abs(tup[0][0]-tup[0][1])), 3)
res.append(wrdv)
return res
#get weighted rating-difference value (wrdv)
def get_wrdv(arr, gamma=1.5, alpha=1.5, b=1):
res=[]
for tup in arr:
r_u = tup[0][0]
r_v = tup[0][1]
x_mean = tup[1][0]
x_sd = tup[1][1]
wrdv_numerator = (1+abs(r_u-r_v))*((1+abs(r_u-r_v))**.5)
wrdv_denominator = ((alpha*(gamma+(r_v-x_mean)**2)**.5)*(gamma+(r_u-x_mean)**2)**.5)**.5
wrdv = round(wrdv_numerator/wrdv_denominator, 3)
res.append(wrdv)
return res
from math import exp
def get_wrdv2(arr, gamma=.35, alpha=5, b=1):
res=[]
for tup in arr:
r_u = tup[0][0]
r_v = tup[0][1]
mu_x = tup[1][0]
sd_x = tup[1][1]
user_diff = (.5+abs(r_u-r_v))
if r_u>mu_x+b and r_v>mu_x+b:
w=-1
elif r_u<mu_x-b and r_v<mu_x-b:
w=-1
else:
w=0
wrdv = user_diff*(gamma*exp(w))
wrdv = round(exp(wrdv)/(alpha+exp(wrdv)), 3)
res.append(1/wrdv)
return res
def get_wrdv3(arr):
res=[]
for tup in arr:
r_u = tup[0][0]
r_v = tup[0][1]
mu_x = tup[1][0]
sd_x = tup[1][1]
if float(abs(r_u-r_v))==0:
res.append(1.0)
elif float(abs(r_u-r_v))==0.5:
res.append(.95)
elif float(abs(r_u-r_v))==1:
res.append( 0.85)
elif float(abs(r_u-r_v))==1.5:
res.append( 0.65)
elif float(abs(r_u-r_v))==2.0:
res.append( 0.45)
elif float(abs(r_u-r_v))==2.5:
res.append( 0.3)
elif float(abs(r_u-r_v))==3.0:
res.append( 0.24)
elif float(abs(r_u-r_v))==3.5:
res.append( 0.17)
elif float(abs(r_u-r_v))==4.0:
res.append( 0.1)
elif float(abs(r_u-r_v))==4.5:
res.append(.05)
return res
#get probability that
def get_prob(arr):
res=[]
for tup in arr:
wrdv = round((1+abs(tup[0][0]-tup[0][1]))/(1+tup[1]), 3)
res.append(wrdv)
return res
import scipy.stats as ss
import math
def prob_pairs(theta):
def _prob_pairs(v):
prob = float(ss.norm(v[0], v[1]).cdf(theta)) #probability difference < theta (where theta is the threshold param)
#if variance==0 then hack a probability estimate with logistic func if the mean <=1
if math.isnan(prob):
if v[0]<=theta:
prob=1-(math.exp(v[1])/(1000+math.exp(v[1])))
return ((v[0], v[1]), round(prob, 3))
else:
prob=0
return ((v[0], v[1]), prob)
return ((v[0], v[1]), round(prob, 3))
return _prob_pairs
###Final User-Movie Network RDD
def rating_rank(v):
if float(v[1])>4.5: #rating of 5
return (v[0], 1.0)
elif float(v[1])>4.0:
return (v[0], .95)
elif float(v[1])>3.5:
return (v[0], 0.85)
elif float(v[1])>3.0:
return (v[0], 0.75)
elif float(v[1])>2.5:
return (v[0], 0.55)
elif float(v[1])>2.0:
return (v[0], 0.4)
elif float(v[1])>1.5:
return (v[0], 0.3)
elif float(v[1])>1.0:
return (v[0], 0.2)
elif float(v[1])>0.5:
return (v[0], 0.1)
elif float(v[1])>0.0:
return (v[0], .05)
#______________________________________________________________________________#
#INITIALIZE
rdd=rdd.filter(lambda x: x[0]!='user_id')
rt = rdd.map(lambda x: x.split(","))
rt = rt.map(lambda x: x[0:3])
#______________________________________________________________________________#
#USER NETWORK
#(user, (movie, rating)
u_rt2 = rt.map(lambda x: (x[0], (x[1], x[2])))
#(movie, (user, rating)
u_rt3 = u_rt2.map(lambda x: (x[1][0], (x[0], float(x[1][1]))))
#sd5b:(u'4999', (4.0, (u'657', 4.0)))]
#sd5: (u'4999', (0, (u'657', 4.0)))
#standard deviation of users
u_sd1 = u_rt3.mapValues(lambda x: x[1])
u_sd2 = u_sd1.combineByKey(li, app, ext)
u_sd3 = u_sd2.map(sd1)
u_sd3b = u_sd3.map(mean)
u_sd3c = u_sd3b.map(normalize)
#u_sd4 = u_sd3b.map(lambda x: (x[0], (x[1], x[0][1])))
u_sd5 = u_sd3b.join(u_rt3)
u_sd6 = u_sd5.map(lambda x: ((x[0], x[1][0]), x[1][1]))
#u_rt5 schema: ((movie, (movie_mean, movie_stdev)), (userA, ratingA), (userB, ratingB))
u_rt4 = u_sd6.join(u_sd6)
u_rt5 = u_rt4.filter(lambda x: int(x[1][0][0])<int(x[1][1][0]))
# u_rt6 schema: ((userA, userB), (ratingA, ratingB), (movie_mean, movie_stdev))
u_rt6 = u_rt5.map(lambda x: ((x[1][0][0],x[1][1][0]), ((x[1][0][1], x[1][1][1]), x[0][1])))
#rdd user_pairs object schema:
#((userA, userB), ([((userA_rating1, userB_rating_1), (movie1_mean, movie1_stdev),
#((userA_rating_2, userB_rating_2), movie2_stdev),…, ((userA_rating_n, userB_rating_n), movie_n_stdev)]))
user_pairs1 = u_rt6.combineByKey(li, app, ext)
#normalize ratings
user_pairs1b = user_pairs1.mapValues(lambda x:
#user_pairs2 schema: ((userA, userB), [wrdv1, wrdv_2,...,wrdv_n])
user_pairs2 = user_pairs1.mapValues(get_wrdv)
user_pairs3 = user_pairs2.map(sd1)
#user_pairs4 schema: (((userA, userB), (avg_wrdv, sd_wrdv), num_shared_movies))
user_pairs4 = user_pairs3.map(lambda x: (x[0][0], (round(st.mean(x[0][1]), 3), x[1]), len(x[0][1])))
#m_cdf_pairs schema: ((userA, userB), (avg_mag_diffAB, sd_diffAB), sample_size), probability_diff<theta)
u_cdf_pairs = user_pairs4.mapValues(prob_pairs(1))
#Add 'u' to each movie ID for 'user'
#USER_NETWORK schema: ((userA, 'u'), ((userB, 'u'), probability_wrdv<theta)
USER_NETWORK = u_cdf_pairs.map(lambda x: ((x[0][0], 'u'), ((x[0][1], 'u'), x[1][1])))
'''IF USING WRDV2'''
#user_pairs2 schema: ((userA, userB), [wrdv1, wrdv_2,...,wrdv_n])
user_pairs2 = user_pairs1.mapValues(get_wrdv3)
user_pairs3 = user_pairs2.map(sd1)
#user_pairs4 schema: (((userA, userB), (avg_wrdv, sd_wrdv), num_shared_movies))
user_pairs4 = user_pairs3.map(lambda x: (x[0][0], (round(st.mean(x[0][1]), 3), x[1]), len(x[0][1])))
USER_NETWORK = user_pairs4.map(lambda x: ((x[0][0], 'u'), ((x[0][1], 'u'), x[1][0])))
#______________________________________________________________________________#
#MOVIE NETWORK
#(movie, (user, rating)
m_rt2 = rt.map(lambda x: (x[1], (x[0], x[2])))
#(user, (movie, rating)
m_rt3 = m_rt2.map(lambda x: (x[1][0], (x[0], float(x[1][1]))))
#standard deviation of users
m_sd1 = m_rt3.mapValues(lambda x: x[1])
m_sd2 = m_sd1.combineByKey(li, app, ext)
m_sd3 = m_sd2.map(sd1)
m_sd3b = m_sd3.map(mean)
#u_sd4 = u_sd3b.map(lambda x: (x[0], (x[1], x[0][1])))
m_sd5 = m_sd3b.join(m_rt3)
m_sd6 = m_sd5.map(lambda x: ((x[0], x[1][0]), x[1][1]))
'''
m_sd4 = m_sd3.map(lambda x: (x[0][0], (x[1]), x[0][1]))
m_sd5 = m_sd4.join(m_rt3)
m_sd6 = m_sd5.map(lambda x: ((x[0], x[1][0]), x[1][1]))
'''
#m_rt5 schema: ((user, user_stdev), (movieA, ratingA), (movieB, ratingB))
m_rt4 = m_sd6.join(m_sd6)
m_rt5 = m_rt4.filter(lambda x: int(x[1][0][0])<int(x[1][1][0]))
# ((movieA, movieB), (ratingA, ratingB), user_stdev)
m_rt6 = m_rt5.map(lambda x: ((x[1][0][0],x[1][1][0]), ((x[1][0][1], x[1][1][1]), x[0][1])))
#rdd user_pairs object schema:
#[(movieA, movieB), ([((movieA_rating_1, movieB_rating_1), user1_stdev),
#((movieA_rating_2, movieB_rating_2), user2_stdev),…, ((movieA_rating_n, movieB_rating_n), usern_stdev)]))
movie_pairs1 = m_rt6.combineByKey(li, app, ext)
#movie_pairs2 schema: ((movieA, movieB), [weighted_rating_difference_value_1, weighted_rating_difference_value_2,...,weighted_rating_difference_value_n])
movie_pairs2 = movie_pairs1.mapValues(get_wrdv)
movie_pairs3 = movie_pairs2.map(sd1)
#movie_pairs4 schema: (((movieA, movieB), (avg_wrdv, sd_wrdv), num_shared_users))
movie_pairs4 = movie_pairs3.map(lambda x: (x[0][0], (round(st.mean(x[0][1]), 3), x[1]), len(x[0][1])))
#m_cdf_pairs schema: ((movieA, 'm'), (movieB, 'm')), (avg_mag_diffAB, sd_diffAB), sample_size), probability_diff>theta)
m_cdf_pairs = movie_pairs4.mapValues(prob_pairs(1))
#Add 'm' to each movie ID for 'movie'
#MOVIE_NETWORK schema: ((movieA, 'm'), ((movieB, 'm'), probability_wrdv>theta)
MOVIE_NETWORK = m_cdf_pairs.map(lambda x: ((x[0][0], 'm'), ((x[0][1], 'm'), x[1][1])))
'''IF USING WRDV2'''
#movie_pairs2 schema: ((movieA, movieB), [wrdv1, wrdv_2,...,wrdv_n])
movie_pairs2 = movie_pairs1.mapValues(get_wrdv3)
movie_pairs3 = movie_pairs2.map(sd1)
#movie_pairs4 schema: (((movieA, movieB), (avg_wrdv, sd_wrdv), num_shared_movies))
movie_pairs4 = movie_pairs3.map(lambda x: (x[0][0], (round(st.mean(x[0][1]), 3), x[1]), len(x[0][1])))
MOVIE_NETWORK = movie_pairs4.map(lambda x: ((x[0][0], 'm'), ((x[0][1], 'm'), x[1][0])))
#______________________________________________________________________________#
#USER-MOVIE NETWORK
um_ratings = u_rt2.mapValues(rating_rank)
um_ratings2 = um_ratings.map(lambda x: ((x[0], 'u'), ((x[1][0], 'm'), x[1][1])))
#um_ratings2 = u_rt2.map(lambda x: ((x[0], 'u'), ((x[1][0], 'm'), float(x[1][1]))))
user_movie_network0 = USER_NETWORK.union(MOVIE_NETWORK)
USER_MOVIE_NETWORK_Gaussian = user_movie_network0.union(um_ratings2)