Impact of Learning Data Statistics on the Performance of a Recommendation System Based on MovieLens Data
摘要
Recommendation systems are an efficient way to improve user satisfaction on the internet and personalise web services. They use various algorithms to analyse large amounts of data and offer customers the most useful products. The accuracy of the offered propositions is the most important feature of all recommendation systems, which is closely related to the principle of the methods. However, the input data also influences their performance. The objective of this study is to determine whether there is a correlation between data statistics, such as density, shape and skewness, and the performance of the algorithm as measured by RMSE. Additionally, we examined the coverage of both items and users with respect to the input data. The verification was performed using classical item and user-based methods with cosine similarity.