Off-line quality improvement approaches based on loss functions

In this thesis, quality improvement approaches under gamma distributed noise variables has been studied. Within the scope of proposed approaches, firstly, we investigate the effects of this skewed structure on responses and based upon this scenario the proposed density function under gamma distributed noise variable is obtained. Furthermore, a new asymmetric loss function is introduced and its properties have been investigated. Finally, considering the proposed loss function, a new dual response surface approach is given under responses have proposed probability density function which is obtained in this thesis. A simulated experimental design is conducted under the assumption that gamma distributed noise variables get values randomly during the experiment and the relevance between the structure of theoretical pdf and forms of distribution of simulated responses at experiment runs is illustrated. The losses for all experimental runs are identified using proposed loss function and also an application of proposed dual response approach is performed.