Marappan R* and Sarveshvara Raja
Senior Assistant Professor, School of Computing, India
*Corresponding author: Marappan R, School of Computing, SASTRA Deemed University, Thanjavur, India
Submission: June 16, 2022;Published: September 27, 2022
ISSN:2832-4463 Volume2 Issue3
Recommender systems are playing a major role in the information filtering to the users. Recently there are different recommender models developed for various artificial intelligence, machine learning and data analytics applications. This research focuses on the types of movie recommender systems and building a collaborative filtering model using Python.
Keywords: Recommender systems; Information filtering; Collaborative filtering; Machine learning; Data analytics; Artificial intelligences
Recommender systems are filtering the required information to the users based on their
interests, styles of learning, and other primary characteristics. The recommender systems
such as Collaborative Filtering (CF) are also developed for movie applications [1,2]. The
recommendation systems are implemented by the companies for the following main criterion:
A. Improving the retention: The users choices are continuously catered to make loyal
customers
B. Increasing the sales: To increase the sales 10% to 50% due to accurate “You must
also like” recommendations of the products.
C. Form habits: To provide accuracy in the contents results in developing strong habits,
customers usage patterns.
D. Accelerate work: Time saving up to 80% due to specific recommendations and to
support the research work.
Figure 1:Types of movie recommenders.
The types of movie recommenders systems are sketched in (Figure 1) with CF and content-based filtering [3,4]. Companies such as Amazon, Pandora, and Netflix are using analytics to predict customer behaviors and provide recommendations. The companies such as Amazon, Pandora, Twitter, and Netflix track and how they use this data are sketched in (Figure 2-5).
Figure 2:Amazon recommender.
Figure 3:Pandora recommender.
Figure 4:Twitter recommender.
Figure 5:Netflix recommender.
Figure 6:CF & limitations.
Limitations of user CF
a. One can watch specific movies that no one else watches
which results in no recommendation.
b. Not enough ratings to match new movies.
c. For new users, the user is not rated and watched many
movies and it results in users mapping problems.
Building a user CF in python
The user CF model is developed in Python with the following
steps:
a. Install and import the surprise package.
b. Load the instance ml-100k.
c. Split the data into training and testing.
d. Apply the KNN algorithm to the training set.
e. Display the 25 best neighbors to the user with ratings.
f. Predict the rating for a user not yet watched the movie.
g. Find the expected ratings.
h. Test using the testing dataset for every movie.
The Python implementation is as follows: (Figure 7)
Figure 7:The Python implementation.
This research analyzed the types of movie recommender systems and developed a CF model using Python for movie applications. In the future, different soft computing strategies can be used for the implementation to obtain better accuracy [5-10].
© 2022 Marappan R. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and build upon your work non-commercially.