Application of Linear Algebra in the Data Mining: A Proposal of New Matrix Decompositions

In the datafication era, huge amount of data can be collected and then analyzed. In all the contexts, very large data sets may be analyzed not only by using statistical approach, but may be viewed as very large size matrices. The linear algebra approach may be very useful, for example, in the data mining phase, with the advantage that all the information is maintained in the data set. In this work, two new matrix decompositions are presented. Given the characteristics of the new decompositions, their use in the data analysis may be certainly very attractive for the practitioners.