* {userId: {movieId: rating}}
* {movieId: {userId: rating}}
"""
urm = {}
movie_ratings = {}
with open('resources/train.csv', 'r') as urm_raw:
reader = csv.reader(urm_raw)
# create a nested dict; {userId: {movieId: rating}}
for row in reader:
if row[0] != 'userId':
urm.setdefault(int(row[0]), {}).setdefault(int(row[1]))
urm[int(row[0])][int(row[1])] = float(row[2])
movie_ratings.setdefault(int(row[1]), {}).setdefault(int(row[0]))
movie_ratings[int(row[1])][int(row[0])] = float(row[2])
print(top_n_match(movie_ratings, 1, skr=10, similarity=cosine_sim))
time = datetime.now()
movie_knn = {}
user_knn = {}
"""
with open('resources/test.csv') as test_raw:
reader = csv.reader(test_raw)
# create a nested dict for test user
for row in reader:
if row[0] != 'userId':
user_knn[int(row[0])] = top_n_match(urm, int(row[0]), skr=10, n=20, similarity=cosine_sim)
print('fatto user: ' + row[0])
"""
for movie in movie_ratings:
movie_knn.setdefault(movie, {})
icm = {}
with open('resources/icm.csv', 'r') as urm_raw:
reader = csv.reader(urm_raw)
# create a nested dict; {userId: {movieId: rating}}
for row in reader:
if row[0] != 'itemId':
icm.setdefault(int(row[0]), {}).setdefault(int(row[1]))
icm[int(row[0])][int(row[1])] = 1
print('computing right now')
time = datetime.datetime.now()
print(datetime.datetime.now() - time)
knn = {}
time = datetime.datetime.now()
"""
for movie in sorted(icm):
knn[movie] = top_n_match(icm, movie, skr=5, n=25, similarity=item_sim)
print('fatto movie: ' + str(movie))
print(len(knn))
"""
with open('resources/knn_item_movie_25.csv', 'w') as knn_raw:
fieldnames = ['itemId', 'neighborId', 'similarity']
w = csv.DictWriter(knn_raw, fieldnames=fieldnames)
w.writeheader()
for movie in sorted(icm):
knn[movie] = top_n_match(icm, movie, skr=5, n=25, similarity=item_sim)
print('fatto movie: ' + str(movie))
for other in knn[movie]:
knn_raw.write(str(movie) + ',' + str(other) + ',' + str(knn[movie][other]) + '\n')
print(datetime.datetime.now() - time)
urm = {}
with open('resources/train.csv', 'r') as urm_raw:
reader = csv.reader(urm_raw)
# create a nested dict; {userId: {movieId: rating}}
for row in reader:
if row[0] != 'userId':
urm.setdefault(int(row[1]), {}).setdefault(int(row[0]))
urm[int(row[1])][int(row[0])] = float(row[2]) + item_bias[int(row[1])] + user_bias[int(row[0])]
print('computing right now')
time = datetime.datetime.now()
print(datetime.datetime.now() - time)
knn = {}
time = datetime.datetime.now()
"""
for movie in sorted(icm):
knn[movie] = top_n_match(icm, movie, skr=5, n=25, similarity=item_sim)
print('fatto movie: ' + str(movie))
print(len(knn))
"""
with open('resources/knn_movie_urm_25.csv', 'w') as knn_raw:
fieldnames = ['itemId', 'neighborId', 'similarity']
w = csv.DictWriter(knn_raw, fieldnames=fieldnames)
w.writeheader()
for movie in sorted(urm):
knn[movie] = top_n_match(urm, movie, skr=5, n=25, similarity=cosine_sim)
print('fatto movie: ' + str(movie))
for other in knn[movie]:
knn_raw.write(str(movie) + ',' + str(other) + ',' + str(knn[movie][other]) + '\n')
print(datetime.datetime.now() - time)
rec = sorted(rankings, key=lambda x: -x[0])[0:5]
return rec
urm = {}
with open('resources/train.csv', 'r') as urm_raw:
reader = csv.reader(urm_raw)
# create a nested dict; {userId: {movieId: rating}}
for row in reader:
if row[0] != 'userId':
urm.setdefault(int(row[0]), {}).setdefault(int(row[1]))
urm[int(row[0])][int(row[1])] = float(row[2])
print('computing right now')
time = datetime.datetime.now()
print(top_n_match(urm, 4, 7, similarity=similarity_pearson))
print(datetime.datetime.now() - time)
test = {}
time = datetime.datetime.now()
with open('resources/test.csv') as test_raw:
reader = csv.reader(test_raw)
# create a nested dict for test user
for row in reader:
if row[0] != 'userId':
test[int(row[0])] = top_n_match(urm, int(row[0]), skr=10, similarity=cosine_sim)
print('fatto user: ' + row[0])
print(datetime.datetime.now() - time)
print(len(test))
"""