Recommender Systems using Collaborative Filtering

Abstract

With the vast amount of data that the world has nowadays, institutions are looking for more and more accurate ways of using this data. Companies like Amazon use their huge amounts of data to give recommendations for users. Based on similarities among items, systems can give predictions for a new items rating. Recommender systems use the user, item, and ratings information to predict how other users will like a particular item. Recommender systems are now pervasive and seek to make prot out of cus- tomers or successfully meet their needs. However, to reach this goal, systems need to parse a lot of data and collect information, sometimes from dierent resources, and predict how the user will like the product or item. The computation power needed is considerable. Also, companies try to avoid ooding customer mailboxes with hundreds of products each morning, thus they are looking for one email or text that will make the customer look and act. The motivation for this project comes from the eagerness to get a deep un- derstanding of recommender systems. One of the goals set for this project was to apply machine learning dynamically and to verify the results. Thus, a large dataset is used to test the algorithm and to compare each algorithm in terms of error rate. In this project, a website has been developed that uses dierent techniques for recommendations namely User-based Collaborative Filtering, Item-Based Collab- orative Filtering and Model Based Collaborative Filtering. Every technique has its way of predicting the user rating for a new item based on existing users data. To evaluate each method, I have used Movie Lens, an external data set of users, items, and ratings, and calculated the error rate using Mean Absolute Error Rate (MAE) and Root Mean Squared Error (RMSE)