Performance Assessment of Various Machine Learning Algorithms in Recommendation

Recommendation system has become an inevitable tool for businesses over the years. Its significance is widely recognized for both products as well as services. This study offers a thorough examination of several machine learning algorithms appropriate for recommendation systems designed for diverse domains, such as music and movies. Although there are several algorithms available for creating suggestions, some jobs may benefit from the use of a particular method. This article examines a number of basic and sophisticated algorithms used in recommendation systems, explains their applications, and analyses their advantages and disadvantages. This research compares the implementation of movie recommendation using single value decomposition plus-plus (SVD++) with popular machine learning techniques like k-nearest neighbor (K-NN) and singular value decomposition (SVD). Using the MovieLens 100 K and 1M datasets, it is experimentally proven by measurement of the Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). The outcome demonstrates that the SVD++ provides a lower error rate.