Metodología de superficie de respuesta y Modelo de regresión logística: Aplicación a conjuntos de datos reales sobre la fabricación de circuitos integrados
Metodología de superficie de respuesta y Modelo de regresión logística: Aplicación a conjuntos de datos reales sobre la fabricación de circuitos integrados
-
DOI: https://doi.org/10.22533/at.ed.1232517102
-
Palavras-chave: Diseño y análisis de experimentos, Modelos, Metodología superficie de respuesta y Pruebas de hipótesis
-
Keywords: Design and analysis of experiments, Models, Response surface methodology and Hypothesis testing
-
Abstract: The Response Surface Methodology (MSR) can be defined as a strategy that encompasses the following points: choosing an experimental design that allows to adequately measure the behavior of the response of interest; determine a model that describes the behavior of the data obtained through the experimental design, which implies doing some statistical tests to verify if the model is adequate. Once we have an adequate model, we proceed to find the combination of the levels of the input factors that produce the optimal response. MSR had its origin in the 1930s in work by Sisar, Yates, and others, however, it was formally developed by Box and Wilson (1951). The objective of this work is to describe three applications of MSR to real data sets: the first is an experiment carried out in a Mexican electronics industry; whose objective is to reduce the number of electronic devices that break at a certain stage of their manufacturing process, due to the sudden changes in temperature that occur there, the problem is that some wafers do not resist these changes in temperature and break (Castro Montoya, 1995). A silicon wafer is an electronic device in which microcircuits are integrated to be processed together, that is, the wafer is the medium that allows hundreds of dice or chips to be processed at the same time, which must meet certain electrical properties.
- René Castro Montoya
- José Vidal Jiménez Ramírez