Premium
New developments for net‐effect plots
Author(s) -
Zhang Xin
Publication year - 2013
Publication title -
wiley interdisciplinary reviews: computational statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.693
H-Index - 38
eISSN - 1939-0068
pISSN - 1939-5108
DOI - 10.1002/wics.1247
Subject(s) - statistical graphics , graphics , bivariate analysis , computer science , dimensionality reduction , subspace topology , dimension (graph theory) , net (polyhedron) , visualization , data mining , sliced inverse regression , graphical model , reduction (mathematics) , algorithm , theoretical computer science , mathematics , artificial intelligence , machine learning , computer graphics (images) , combinatorics , geometry
We review the graphics for studying the net‐effects of predictors, including both the global and local net‐effect plots. Then some new definitions of net‐effects and corresponding graphical methods are introduced for studying and visualizing the main and interaction net‐effects of the mean functions and the distribution functions. A sufficient dimension reduction method, called central solution subspace bivariate sliced inverse regression (CSS‐BiSIR), is proposed for reducing the size of the graphical problem. This facilitates the graphical interpretations of the net‐effects, and also allows us to visualize the net‐effects of nonelliptically distributed predictors. WIREs Comput Stat 2013, 5:105–113. doi: 10.1002/wics.1247 This article is categorized under: Statistical and Graphical Methods of Data Analysis > Dimension Reduction Statistical and Graphical Methods of Data Analysis > Statistical Graphics and Visualization