-
Notifications
You must be signed in to change notification settings - Fork 2
/
test.py
245 lines (199 loc) · 9.24 KB
/
test.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
from sklearn.neighbors import NearestNeighbors
from scipy.sparse import csr_matrix
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
# datasets books,student,rating
books = pd.read_csv('books.csv', error_bad_lines=False, encoding="latin-1")
# columns of books dataset
books.columns = ['ISBN', 'bookTitle',
'bookAuthor', 'yearOfPublication', 'publisher']
print(books)
users = pd.read_csv('student.csv', error_bad_lines=False, encoding="latin-1")
# columns of student dataset
users.columns = ['userID', 'Name', 'Age', 'Interest']
print(users)
ratings = pd.read_csv('ratings.csv',
error_bad_lines=False, encoding="latin-1")
# columns of rating dataset
ratings.columns = ['userID', 'ISBN', 'bookRating']
print(ratings)
# shape of rating dataset,gives the dimension of dataset i.e the number of rows and columns
print(ratings.shape)
# list of rating columns
print(list(ratings.columns))
print(books.shape)
# list of books columns
print(list(books.columns))
print(users.shape)
# list of student columns
print(list(users.columns))
# rating distributution using histogram
plt.rc("font", size=15)
ratings.bookRating.value_counts(sort=False).plot(kind='bar')
plt.title('Rating Distribution\n')
plt.xlabel('Rating')
plt.ylabel('Count')
plt.savefig('system1.png', bbox_inches='tight')
plt.show()
# student age distributution using histogram
users.Age.hist(bins=[18, 20, 22, 24, 26, 28, 30, 32, 40])
plt.title('Age Distribution\n')
plt.xlabel('Age')
plt.ylabel('Count')
plt.savefig('system2.png', bbox_inches='tight')
plt.show()
# recommendation based on rating count
rating_count = pd.DataFrame(ratings.groupby('ISBN')['bookRating'].count())
print(rating_count)
# sorting of the counts of rating to get the highest rated books
rating_count.sort_values('bookRating', ascending=False).head()
xy = rating_count.sort_values('bookRating', ascending=False).head(5)
print(xy)
# books details of first 5 book which received highest rating by students
most_rated_books = pd.DataFrame(['978-8120349391', '978-0198070887', '978-9351341741',
'978-0198083542', '978-9351343257'], index=np.arange(5), columns=['ISBN'])
most_rated_books_summary = pd.merge(most_rated_books, books, on='ISBN')
most_rated_books_summary
# recommendations based on correlations
# here Pearson correlation coefficient used to measure the linear correlation between
# two variable --- the ratings for two books
# fetch the average rating and the count of rating each book received
average_rating = pd.DataFrame(ratings.groupby('ISBN')['bookRating'].mean())
print(average_rating)
average_rating['ratingCount'] = pd.DataFrame(
ratings.groupby('ISBN')['bookRating'].count())
average_rating.sort_values('ratingCount', ascending=False).head(10)
# here the main disadvantage is that, the book which got highest number of rating
# has the rating average is low
# observation-- in this dataset the book that received the most rating counts was not highly
# rated at all. So if we are going to use recommendation based on
# rating counts,we would definitely make mistake or wrong recommendation.
# to ensure statistical significance,
# student who rate books and their count >=3
counts1 = ratings['userID'].value_counts()
print(counts1)
ratings = ratings[ratings['userID'].isin(counts1[counts1 >= 1].index)]
print(ratings)
counts = ratings['bookRating'].value_counts()
print(counts)
# rating of book > 2
ratings = ratings[ratings['bookRating'].isin(counts[counts >= 1].index)]
print(ratings)
# ---------------------------------------------------------------
# using pearson correlation
# rating matrix
# convert the rating table into 2D matrix.
# generate sparse matrix because not all students rated book
# by using pivot table we will be able to create combination of userId and isbn
# this will give us that whether the student is rated a book either NAN is given
#
ratings_pivot = ratings.pivot(index='userID', columns='ISBN').bookRating
userID = ratings_pivot.index
print(userID)
ISBN = ratings_pivot.columns
print(ISBN)
print(ratings_pivot.shape)
ratings_pivot.head()
# pearson algorithm to find correlation between the isbn with other
someBookIsbn_ratings = ratings_pivot['978-0070634244']
similar_to_someBookIsbn_ratings = ratings_pivot.corrwith(someBookIsbn_ratings)
corr_someBookIsbn = pd.DataFrame(
similar_to_someBookIsbn_ratings, columns=['pearsonR'])
corr_someBookIsbn.dropna(inplace=True)
corr_summary = corr_someBookIsbn.join(average_rating['ratingCount'])
corr_summary[corr_summary['ratingCount'] >= 2].sort_values(
'pearsonR', ascending=False).head(10)
# book details
books_corr_to_someBookIsbn = pd.DataFrame(['978-0070634244', '978-9351341741'],
index=np.arange(2), columns=['ISBN'])
corr_books = pd.merge(books_corr_to_someBookIsbn, books, on='ISBN')
corr_books
# ----------------------------------------------------------------------------------
# now recommend book using KNN algorithm
# Collaborative Filtering Using k-Nearest Neighbors (kNN)
# kNN is a machine learning algorithm to find clusters of similar users based on common book ratings,
# and make predictions using the average rating of top-k nearest neighbors.
# For example, we first present ratings in a matrix with the matrix having one row for each item (book) and one column for each user,
# merging of rating abd books dataset based on ISBN
combine_book_rating = pd.merge(ratings, books, on='ISBN')
columns = ['yearOfPublication', 'publisher']
# dropping these columns because these are not required
print(columns)
combine_book_rating = combine_book_rating.drop(columns, axis=1)
combine_book_rating.head(20)
# We then group by book titles and create a new column for total rating count.
combine_book_rating = combine_book_rating.dropna(axis=0, subset=['bookTitle'])
book_ratingCount = (combine_book_rating.
groupby(by=['bookTitle'])['bookRating'].
count().
reset_index().
rename(columns={'bookRating': 'totalRatingCount'})
[['bookTitle', 'totalRatingCount']]
)
book_ratingCount.head(10)
# We combine the rating data with the total rating count data,
# this gives us exactly what we need to find out which books are popular and filter out lesser-known books.
rating_with_totalRatingCount = combine_book_rating.merge(
book_ratingCount, left_on='bookTitle', right_on='bookTitle', how='left')
rating_with_totalRatingCount.head(10)
pd.set_option('display.float_format', lambda x: '%.3f' % x)
print(book_ratingCount['totalRatingCount'].describe())
print(book_ratingCount['totalRatingCount'].quantile(np.arange(.9, 1, .01)))
# threshold value = 5 that is
# if this is greater than totalRating then suggest
popularity_threshold = 5
rating_popular_book = rating_with_totalRatingCount.query(
'totalRatingCount >= @popularity_threshold')
rating_popular_book.head(20)
rating_popular_book.shape
# popular book with highest rating count
print(rating_popular_book)
combined = rating_popular_book.merge(
users, left_on='userID', right_on='userID', how='left')
print(combined)
# recommend based on user interest lets take Networks and
interest_user_rating = combined[combined['Interest'].str.contains(
"Networks")]
interest_user_rating = interest_user_rating .drop('Age', axis=1)
interest_user_rating.head(50)
# now time to apply cosine similarity
# in this each book is represented in vector
'''
Implementing kNN
We convert our table to a 2D matrix, and fill the missing values with zeros
(since we will calculate distances between rating vectors). We then transform the values(ratings)
of the matrix dataframe into a scipy sparse matrix for more efficient calculations.
Finding the Nearest Neighbors We use unsupervised algorithms with sklearn.neighbors.
The algorithm we use to compute the nearest neighbors is “brute”, and we specify “metric=cosine”
algorithm will calculate the cosine similarity between rating vectors. Finally, we fit the model.
'''
interest_user_rating = interest_user_rating.drop_duplicates(
['userID', 'bookTitle'])
print(interest_user_rating)
interest_user_rating_pivot = interest_user_rating.pivot(
index='bookTitle', columns='userID', values='bookRating').fillna(0)
print(interest_user_rating_pivot)
# csr_matrix is sparse matrix
interest_user_rating_matrix = csr_matrix(interest_user_rating_pivot.values)
print(interest_user_rating_matrix)
print(interest_user_rating)
# implementation of KNN
model_knn = NearestNeighbors(metric='cosine', algorithm='brute')
model_knn.fit(interest_user_rating_matrix)
print(model_knn)
query_index = np.random.choice(interest_user_rating_pivot.shape[0])
print(query_index)
distances, indices = model_knn.kneighbors(
interest_user_rating_pivot.iloc[query_index, :].values.reshape(1, -1), n_neighbors=4)
interest_user_rating_pivot.iloc[query_index, :].values.reshape(1, -1)
print(interest_user_rating_pivot)
print("Recommendation for the book:- ",
interest_user_rating_pivot.index[query_index])
for i in range(0, len(distances.flatten())):
if i == 0:
print('Recommendations for {0}:\n'.format(
interest_user_rating_pivot.index[query_index]))
else:
print('{0}: {1}, with distance of {2}:'.format(
i, interest_user_rating_pivot.index[indices.flatten()[i]], distances.flatten()[i]))