A Comparative Study of Collaborative Filtering Recommendation Systems Using Algorithms to Impute Large Sparse Matrices
In the modern era of computing, recommendation systems are a key component for enterprise systems and consumer applications including e-commerce and web applications. The challenge for these systems is the accuracy and quality of the calculations especially when dealing with sparse amounts of data. This research provides an empirical study of the problems associated with a sparse matrix when encountered by collaborative filtering recommendation systems. The research conducts a comparative analysis of different algorithms used to address the issue of sparse data while trying to predict and prescribe (recommend) the optimal choice to users. The research will compare statistical techniques used to impute missing data with other estimations techniques for predicting missing values. The research will show why Matrix Factorization, Maximum Likelihood Estimation (MLE) or Gradient Optimization methods work better for large sparse matrices, over simple mean, sub-group mean or regression methods.
Lindo, Steven Christopher, "A Comparative Study of Collaborative Filtering Recommendation Systems Using Algorithms to Impute Large Sparse Matrices" (2016). ETD Collection for Pace University. AAI10182708.
Remote User: Click Here to Login (must have Pace University remote login ID and password. Once logged in, click on the View More link above)