A new method for design parameter optimization of products or processes with an ordinal categorical response

Download
2022-8
Erdoğan, Pınar
It is an important design problem to obtain the best parameter values that are insensitive to the variability of input, process or environmental factors of a product or process. Therefore, many industrial organizations use robust parameter design (RPD). While there are many methods in the literature for design parameter optimization with continuous quality characteristics, fewer studies are found for ordered categorical quality characteristics. This study proposes a new method for finding robust levels of design parameters of products and processes with ordinal categorical responses. Many approaches use expected value and variance as measures of location and dispersion respectively when the response is categorical. However, these measures used for numerical data are not meaningful to summarize categorical data. The developed method uses the median value and coefficient of ordinal variation (COV) for ordinal categorical data as location and dispersion measures, respectively. In addition, the Extreme Gradient Boosting (XGBoost) algorithm is presented as an alternative to Random Forests and Logistic Regression, which have been used in RPD studies in the literature. The classification performances of these algorithms are compared with Multi-Objective Decision Making methods. Based on this comparison, Random Forests algorithm is used to predict the category probabilities. These probabilities are used to calculate median and COV. Ordinal Logistic Regression and least squares regression are used to model these measures as functions of design parameters. The best parameter values are determined by solving a non-linear optimization problem. The proposed method is applied to different problems, and the results are discussed.

Suggestions

Comparison of methods for robust parameter design of products and processes with an ordered categorical response
Bacanlı Karabulut, Gonca; Köksal, Gülser; Department of Industrial Engineering (2013)
Robust design of products or processes with categorical response has more momentous role in industrial experiments for quality improvements, because ordinal categorical quality characteristics are encountered more frequent than continuous ones in industry. In this study, five optimization methods for an ordered categorical response are compared with each other: Logistic Regression Model Optimization (LRMO), Accumulation Analysis (AA), Weighted Signal-to-noise Ratio (WSNR), Scoring Scheme (SS), and Weighted ...
A REPLICATION APPROACH TO INTERVAL ESTIMATION IN SIMULATION
Köksalan, Mustafa Murat; BASOZ, N (1991-12-11)
The authors modify an earlier approach developed for reducing the bias of the estimator for the mean response in simulation caused by the initial conditions. They try to balance the bias of the estimator in a simulation run by imposing a bias in the opposite direction in a companion run by suitably setting its initial conditions. They present analytical results for the bias of the estimator for AR(1) and M/M/s processes. They suggest making independent replications of the pairs of runs to construct a confid...
A reformulation of the ant colony optimization algorithm for large scale structural optimization
Hasançebi, Oğuzhan; Saka, M.p. (2011-01-01)
This study intends to improve performance of ant colony optimization (ACO) method for structural optimization problems particularly with many design variables or when design variables are chosen from large discrete sets. The algorithm developed with ACO method employs the so-called pheromone scaling approach to overcome entrapment of the search in a poor local optimum and thus to recover efficiency of the method for large-scale optimization problems. Besides, a new formulation is proposed for the local upda...
An entropy based input variable selection approach to identify equally informative subsets for data driven hydrological models
Karakaya, Gülşah; Galelli, Stefano; Ahipaşoğlu, Selin Damla (null; 2015-04-15)
Input Variable Selection (IVS) is an essential step in hydrological modelling problems, since it allows determining the optimal subset of input variables from a large set of candidates to characterize a preselected output. Interestingly, most of the existing IVS algorithms select a single subset, or, at most, one subset of input variables for each cardinality level, thus overlooking the fact that, for a given cardinality, there can be several subsets with similar information content. In this study, we devel...
An Objective Methodology For Merging Satellite And Model Based Soil Moisture Products
Yılmaz, Mustafa Tuğrul; Anderson, Martha; Hain, Chris (2012-04-19)
An objective methodology that does not require any user-defined parameter assumptions is introduced to obtain an improved soil moisture product along with associated uncertainty estimates. This new product is obtained by merging model-, thermal infrared remote sensing-, and microwave remote sensing-based soil moisture estimates in a least squares framework where uncertainty estimates for each product are obtained using triple collocation. The merged anomaly product is validated against in situ based soil mo...
Citation Formats
P. Erdoğan, “A new method for design parameter optimization of products or processes with an ordinal categorical response,” M.S. - Master of Science, Middle East Technical University, 2022.