Recommender Systems

Abstract

In this project, a key domain in the field of information filtering, known as 'recommender systems' or recommendation engines, have been explored and studied. A number of different recommendation algorithms are implemented and documented. The main focus was memory-based collaboration filtering techniques, for prediction, and forecasting. both user-centric, and item-centric collaborative filtering algorithms have been explored, including Pearson Product-moment Correlation Coefficient, Vector, or Cosine Similarity, Euclidean Distance, and Slope One. Some performance evaluation metrics are also implemented like Absolute Mean Error, and running time of the algorithm.