import operator
from decimal import Decimal
from math import sqrt
import numpy as np
from django.db.models import Avg, Count
from django.http import JsonResponse
from analytics.models import Rating
from collector.models import Log
from moviegeeks.models import Movie
from recommender.models import SeededRecs
from recs.bpr_recommender import BPRRecs
from recs.content_based_recommender import ContentBasedRecs
from recs.funksvd_recommender import FunkSVDRecs
from recs.fwls_recommender import FeatureWeightedLinearStacking
from recs.neighborhood_based_recommender import NeighborhoodBasedRecs
from recs.popularity_recommender import PopularityBasedRecs
def get_association_rules_for(request, content_id, take=6):
data = SeededRecs.objects.filter(source=content_id) \
.order_by('-confidence') \
.values('target', 'confidence', 'support')[:take]
return JsonResponse(dict(data=list(data)), safe=False)
def recs_using_association_rules(request, user_id, take=6):
events = Log.objects.filter(user_id=user_id)\
.order_by('created')\
.values_list('content_id', flat=True)\
.distinct()
seeds = set(events[:20])
rules = SeededRecs.objects.filter(source__in=seeds) \
.exclude(target__in=seeds) \
.values('target') \
.annotate(confidence=Avg('confidence')) \
.order_by('-confidence')
recs = [{'id': '{0:07d}'.format(int(rule['target'])),
'confidence': rule['confidence']} for rule in rules]
print("recs from association rules: \n{}".format(recs[:take]))
return JsonResponse(dict(data=list(recs[:take])))
def chart(request, take=10):
sorted_items = PopularityBasedRecs().recommend_items_from_log(take)
ids = [i['content_id'] for i in sorted_items]
ms = {m['movie_id']: m['title'] for m in
Movie.objects.filter(movie_id__in=ids).values('title', 'movie_id')}
if len(ms) > 0:
sorted_items = [{'movie_id': i['content_id'],
'title': ms[i['content_id']]} for i in sorted_items]
else:
print("No data for chart found. This can either be because of missing data, or missing movie data")
sorted_items = []
data = {
'data': sorted_items
}
return JsonResponse(data, safe=False)
def pearson(users, this_user, that_user):
if this_user in users and that_user in users:
this_user_avg = sum(users[this_user].values()) / len(users[this_user].values())
that_user_avg = sum(users[that_user].values()) / len(users[that_user].values())
all_movies = set(users[this_user].keys()) & set(users[that_user].keys())
dividend = 0
a_divisor = 0
b_divisor = 0
for movie in all_movies:
if movie in users[this_user].keys() and movie in users[that_user].keys():
a_nr = users[this_user][movie] - this_user_avg
b_nr = users[that_user][movie] - that_user_avg
dividend += a_nr * b_nr
a_divisor += pow(a_nr, 2)
b_divisor += pow(b_nr, 2)
divisor = Decimal(sqrt(a_divisor) * sqrt(b_divisor))
if divisor != 0:
return dividend / Decimal(sqrt(a_divisor) * sqrt(b_divisor))
return 0
def jaccard(users, this_user, that_user):
if this_user in users and that_user in users:
intersect = set(users[this_user].keys()) & set(users[that_user].keys())
union = set(users[this_user].keys()) | set(users[that_user].keys())
return len(intersect) / Decimal(len(union))
else:
return 0
def similar_users(request, user_id, sim_method):
min = request.GET.get('min', 1)
ratings = Rating.objects.filter(user_id=user_id)
sim_users = Rating.objects.filter(movie_id__in=ratings.values('movie_id')) \
.values('user_id') \
.annotate(intersect=Count('user_id')).filter(intersect__gt=min)
dataset = Rating.objects.filter(user_id__in=sim_users.values('user_id'))
users = {u['user_id']: {} for u in sim_users}
for row in dataset:
if row.user_id in users.keys():
users[row.user_id][row.movie_id] = row.rating
similarity = dict()
switcher = {
'jaccard': jaccard,
'pearson': pearson,
}
for user in sim_users:
func = switcher.get(sim_method, lambda: "nothing")
s = func(users, user_id, user['user_id'])
if s > 0.2:
similarity[user['user_id']] = round(s, 2)
topn = sorted(similarity.items(), key=operator.itemgetter(1), reverse=True)[:10]
data = {
'user_id': user_id,
'num_movies_rated': len(ratings),
'type': sim_method,
'topn': topn,
'similarity': topn,
}
return JsonResponse(data, safe=False)
def similar_content(request, content_id, num=6):
sorted_items = ContentBasedRecs().seeded_rec([content_id], num)
data = {
'source_id': content_id,
'data': sorted_items
}
return JsonResponse(data, safe=False)
def recs_cb(request, user_id, num=6):
sorted_items = ContentBasedRecs().recommend_items(user_id, num)
data = {
'user_id': user_id,
'data': sorted_items
}
return JsonResponse(data, safe=False)
def recs_fwls(request, user_id, num=6):
sorted_items = FeatureWeightedLinearStacking().recommend_items(user_id, num)
data = {
'user_id': user_id,
'data': sorted_items
}
return JsonResponse(data, safe=False)
def recs_funksvd(request, user_id, num=6):
sorted_items = FunkSVDRecs().recommend_items(user_id, num)
data = {
'user_id': user_id,
'data': sorted_items
}
return JsonResponse(data, safe=False)
def recs_bpr(request, user_id, num=6):
sorted_items = BPRRecs().recommend_items(user_id, num)
data = {
'user_id': user_id,
'data': sorted_items
}
return JsonResponse(data, safe=False)
def recs_cf(request, user_id, num=6):
min_sim = request.GET.get('min_sim', 0.1)
sorted_items = NeighborhoodBasedRecs(min_sim=min_sim).recommend_items(user_id, num)
print(f"cf sorted_items is: {sorted_items}")
data = {
'user_id': user_id,
'data': sorted_items
}
return JsonResponse(data, safe=False)
def recs_pop(request, user_id, num=60):
top_num = PopularityBasedRecs().recommend_items(user_id, num)
data = {
'user_id': user_id,
'data': top_num[:num]
}
return JsonResponse(data, safe=False)
def lda2array(lda_vector, len):
vec = np.zeros(len)
for coor in lda_vector:
if coor[0] > 1270:
print("auc")
vec[coor[0]] = coor[1]
return vec
