|Author:||Tsoi, Sze Leong|
|Title:||The inverted bivariate and multivariate loss functions : properties and applications|
Hong Kong Polytechnic University -- Dissertations
|Department:||Department of Applied Mathematics|
|Pages:||xiv, 220 p. : ill. ; 30 cm.|
|Abstract:||The purpose of this research project is to develop a loss function by inverting the Dirichlet distribution. The motive is to explore the possibility of developing loss functions of higher dimension. Spiring (1993), Spiring and Yeung (1998) and Spiring and Leung (2002) developed a series of Inverted Probability Loss Functions (IPLFs), while most of them are univariate. A multivariate IPLF, which in theory, should have the capability of depicting financial loss when more than one quality characteristics are concerned. The Dirichlet distribution is chosen because of its addictiveness of random variables in nature and its ease to include any number of variables. To depict the long term financial loss of manufacturing a product, it is desired to calculate the expected loss by E[L] = Ω L(x)g(x)dx. L(x) is the loss function and g(x) is the conjugate function, which describes the measurement of the quality characteristic which the manufacturer is interested in. This research project focuses on developing a multivariate inverted probability loss function and some discussion will be made about the conjugate function. The prime objective of this research is to develop an inverted multivariate probability loss function. For this purpose, extending the conjugate function into higher dimension will also be discussed. Although it is highly desired that real data can be collected to form the conjugate function, sometimes it is not feasible to do so and simulation may be the solution to this situation. Three common random variate generation methods, namely the conditional approach, acceptance and rejection approach and gamma factors approach, are used to invert the Dirichlet distribution. The major distinction between the three aforementioned approaches lays in the method in obtaining the estimators, respectively, the method of moments, maximum likelihood estimation and a modified percentile matching method. The theory and the procedure underlying the methods in obtaining estimators are discussed in details and the sets of estimates obtained are then evaluated by the Chi-square goodness-of-fit test to compare the acceptability of the three estimation techniques. Cells arrangement is also discussed and verified by Peacock's two-dimensional Kolmogorov-Smirnov goodness-of-fit test, which is known for its conservativeness. Properties and various probable conjugate distributions are examined to compare the univariate IPLF with the multivariate IPLF. At last, the bivariate inverted normal loss function, introduced by Spiring (1993) and the inverted Dirichlet loss function are applied to data collected in solder paste composition and dimensions of fasteners, as an application of the proposed technique in depicting the financial loss due to deviation in production, and ultimately, to show the feasibility of such bivariate loss functions.|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: