A recommender system to predict the behaviour of an e-commerce page visitor

Abstract

The development of machine learning as a branch of Artificial intelligence has been rapidly increasing in recent decades owing to the expansion of big data. The increasing availability of large amounts of data has also created demands for more efficient data analysis. Algorithms that are based on statistical models, can learn patterns and make predictions to improve a variety of applications. A recommender system explores the computational approach designed to predict the choices of a user toward an item, based on an examination of the user’s prior preferences and actions. The technique known as collaborative filtering belongs to recommender systems and aims to make preference recommendations for the unknown preferences of a new set of users by analysing the preferences of a known set. For the study of collaborative filtering, it is necessary to determine the similarity between a group of users and items, which is often associated with a user’s behaviour and the type of the item, in order to make suggestions based on the preferences of similar users. This master's thesis uses a Python recommender system to predict clicks, cart adds, and orders while researching recommender systems, in particular collaborative filtering, to enhance the outcomes.