Construction of Multivariate Histogram with Variable Windows on Continuous Radom Quantities In case of small Sample Size
Prof. Chung Suing Kao
(Department of Mathematics, National Chung Cheng University)
In the situation of developing, studying or manufacturing very expensive objects, the random samples can be hardly available that leads us to a simple random sample of only small size. However, it may still be important that we obtain a viable histogram for the few continuous variable values to show a distribution so that further understanding about the property or nature of the variable can be attained.
The talk will defy the traditional method of getting fixed equal-width histogram, since it requires substantial amount of sample values. Rather, a new method will be shown with detailed illustrative examples on univariate and bivariate cases by utilizing a board randomness principle and the concept of principal components to deal with the few sample values or sample points for construction of a meaningful histogram. The way to treat higher dimension by extending the bivariate case will also be explained. The breadth of applications with the new method to various scientific work will be presented and discussed.
日 期：108年11月20日(星期三) 16:10~17:00