Book-Recommendation-System
During the last few decades, with the rise of YouTube, Amazon, Netflix, and many other such web services, recommender systems have taken more and more place in our lives. From e-commerce (suggest to buyers articles that could interest them) to online advertisement (suggest to users the right contents, matching their preferences), recommender systems are today unavoidable in our daily online journeys. By analyzing the problems with ‘Book Recommendation System’ feature, how we can predict the best recommendation for users according to their items approach.
A recommendation system helps an organization to create loyal customers and build trust by them desired products and services for which they came on your site. The recommendation system today is so powerful that they can handle the new customer too who has visited the site for the first time. They recommend the products which are currently trending or highly rated and they can also recommend the products which bring maximum profit to the company. Providing specific data analysis and prediction to done with this data. The main objective is to built a predictive recommender model, which could help in predicting – how we can predict the best recommendation for users according to their items approach. This would help us in providing better recommendation item to a right specific users.
We've three dataset -
• Book data – (ISBN, Book-Title, Book-Author, Year-Of-Publication, Publisher, Image-URL-S, Image-URL-M, Image-URL-L)
• Users data - (User-ID, Location, Age)
• Ratings data - (User-ID, ISBN, Book-Rating)
The users interaction is very vital role for recommendation. To successful build collaborative filtering model in recommender system, data preparation is important. Beginning with book data – dropping URL features (i.e 'Image-URL-S', 'Image-URL-M', 'Image-URL-L'). We have some extra columns which are not required for our task like image URLs. And we rename the columns of each file as the name of the column contains space and lowercase letters so we will correct as to make it easy to use. The features depict great analysis by feature engineering.
• Book data – (ISBN, Book-Title, Book-Author, Year-Of-Publication, Publisher, Image-URL-S, Image-URL-M, Image-URL-L)
• Users data - (User-ID, Location, Age)
• Ratings data - (User-ID, ISBN, Book-Rating)
A book recommendation system is a type of recommendation system where we have to recommend similar books to the reader based on his interest. We’ve all record and data with three different dataset – Book dataset (ISBN, Book-Title, Book-Author, Year-Of-Publication, Publisher, Image-URL-S, Image-URL-M, Image-URL-L); Users dataset (User-ID, Location, Age); Ratings dataset (User-ID, ISBN, Book-Rating). Providing specific data analysis and prediction to done with this data. The main objective is to build a predictive recommender model, which could help in predicting – how we can predict the best recommendation for users according to their items approach. This would help us in providing better recommendation item to a right specific user.
As the first step, I performed data preparation (i.e data cleaning and feature engineering) and EDA part. The book_data, users_data and ratings_data dataset was an important dataset to gets smaller insights from its. Analysis feature like book_title, book_author, publisher, year_of_publication, age, book_rating were important to get some insights. Some of the null values were present in feature data, we replaced with mean of that particular feature. Deal with mismatch feature like book_title, book_author, year_of_publication, publisher. Only considering age between 5-90 we took users data to analysis and perform recommendation on it.
After data preparation, building recommendation system based on popularity (i.e ratings). These recommendations are usually given to every user irrespective of personal characterization. Merged book_data dataset and ratings_explicit. Considering ISBNs that were explicitly rated for this recommendation system.
The third step was to do Collaborative Filtering - Memory based approach was our first trial on train and test dataset which uses the memory of previous users interactions to compute users similarities based on items they’ve interacted (i.e user-based approach) or compute items similarities based on the users that have interacted with them (i.e item-based approach). Applying cosine similarity to make item-item similarity need to take transpose of matrix. This matrix would help in manage train-test matrix. After all views predictions based on similarity, we find recommendation on it based on score. Model-based collaborative filtering algorithms provide item recommendation by first developing a model of user ratings. We use Latent Factor Model called Singular Value Decomposition (SVD). SVD made dimensionality reduction technique in machine learning. SVD is a matrix factorization technique. It made us reduces the number of features of a dataset by reducing the space dimension from N-dimension to K-dimension (where K<N). SVD uses a matrix structure where each row represents a user, and each column represents an item. The elements of this matrix are the ratings that are given to items by users.
Model evaluation metrics is important to distinguish the best collaborative filtering – either by memory based or model based approach. The memory based approach – Cosine Similarity shows RMSE score for item based CF is 7.99 and for user based CF it shows 7.99. The score is slightly similar. Model based collaborative filtering made it better score with Latent Factor Model called SVD. The score improved to 1.63 for both SVD RMSE and accuracy score.
-
Data Preparation (Data Cleaning and Feature Engineering)
-
Exploratory Data Analysis
• Univariate Analysis on numeric and categorical features
• Analysis from categorical variables
-
Got top ten books as per ratings
-
Collaborative Filtering
• Memory Based approach - Cosine Similarity
• Model Based approach – Singular Value Decomposition (SVD)
- Model Evaluation Metrics
• RMSE score with each collaborative filtering approach
- Getting top recommendation of books and ratings
-
Among top 20 Authors the highest number of books has been hold by Agatha Christie. Agatha Christie is leading at top with more than 600 counts, followed by William Shakespeare.
-
Harlequin has most number of books published, followed by Silhouette.
-
Number of Books published in yearly are between 1950 - 2005.
-
Most of the users are between 30-40 prefer more books and somewhat we can also view between 20-30.
-
As per ratings "Selected Poems" has been rated most followed by "Little Women". The countplot shows users have rated 0 the most, which means they haven't rated books at all.
-
The top 10 books recommendation as per ratings with top "The lovely Bones: A novel" with 707 book ratings. But this are not based on some recommendation system. They are top 10 books as per ratings.
-
As we perform by cosine similarity in recommendation system it gives 7.94 RMSE score and SVD improved score it to 1.63 RSME score by Singular Value Decomposition model (SVD).
-
The evaluation metrics for SVD is best RMSE score for all dataset (i.e users_data, ratings_data and books_data).
-
As model based approach was best to signify and at last we got top 10 recommended books and ratings respectively.