Identification of the multivariable outliers using T2 eclipse chart based on the improved Partial Least Squares regression

DOI

10.2991/icmse-15.2015.177How to use a DOI?

Keywords

Multivariable outliers; Partial Least Squares regression; T2 eclipse chart

Abstract

When there is multi-variables in a sample, some samples which obviously disturb the relationships among variables are called outlier samples. However the presence of an extremely significant outlier sample tends to conceal some other outlier samples, which bringing great challenge to the identification of multivariable outliers. On this basis, a method of identifying the multivariable outliers in T2 eclipse chart based on the improved Partial Least Squares regression (PLSR) is proposed. It is generally known that some outliers samples fail to be identified owing to significantly outlier samples are prone to influence the variance of T2 chart. To solve this problem, a fuzzy variance computing method is put forward. The improved PLSR based T2 chart can well overcome the masking effect in outliers identification.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Yunlian Liu
AU  - Yanhui Xi
AU  - Jianhua Liu
AU  - Tiebin Wu
AU  - Xinjun Li
PY  - 2015/12
DA  - 2015/12
TI  - Identification of the multivariable outliers using T2 eclipse chart based on the improved Partial Least Squares regression
BT  - Proceedings of the 2015 6th International Conference on Manufacturing Science and Engineering
PB  - Atlantis Press
SP  - 976
EP  - 980
SN  - 2352-5401
UR  - https://doi.org/10.2991/icmse-15.2015.177
DO  - 10.2991/icmse-15.2015.177
ID  - Liu2015/12
ER  -