Authors:
Il-Gyo Chong
1
;
Chenbo Zhu
2
and
Yanfeng Wu
3
Affiliations:
1
Samsung Electronics, Korea, Republic of
;
2
Zhejiang University of Technology, China
;
3
Fudan University, China
Keyword(s):
Semiconductor Manufacturing Line, Turn around Time (TAT), Data Mining, Variable Selection, Variable Importance in the Projection (VIP) Scores, Partial Least Squares Regression.
Related
Ontology
Subjects/Areas/Topics:
Data Mining and Business Analytics
;
Industrial Engineering
;
Methodologies and Technologies
;
Operational Research
Abstract:
Variation reduction of Turn Around Time (TAT) in a manufacturing line is one of the important issues for line optimization. In a manufacturing line with many sequential process steps such as semiconductor fabrication, it is not easy to find the root causes of the TAT variation because (1) there might be a big time gap (more than 30 days) between cause and effect, and (2) there are so many machines (or tools) related with a process. The purpose of this paper is to propose a data mining based method to identify the root cause of TAT variation. We also aim to validate the performance of the proposed method through a simulation study.