CMStatistics 2015: Start Registration
View Submission - CMStatistics
B0763
Title: Analysis of big medical motion data using functional LARS Authors:  Jian Qing Shi - Southern Univesity of Science and Technology (China) [presenting]
Yafeng Cheng - Newcastle University (United Kingdom)
Abstract: How to extend the idea of LARS to select functional variables in a scalar-on-function regression model is discussed. Two new stopping rules are proposed. The new algorithm works very efficiently even for the case with very large number of functional variables. We discuss the idea, method, implementation and the result of a comprehensive simulation study. The method has been applied to evaluate the function of upper limbs for patients after stroke by using motion data. The data include scalar response, more than 200 scalar covariates and 500 functional covariates. Different models are compared.