Appendix 2: Product Upgrades Based on Minimum Expected Quality Loss 351
Conclusion
A general quality loss function having a shape parameter is developed
which is applicable to evaluate the expected quality loss for quality charac
teristics such as nominal-the-best, smaller-the-better, and larger-the-better.
Additionally, we present an appropriate range of shape parameter values in
the proposed general quality loss function to accommodate the impacts of
upgrades to fielded systems.
By plotting the loss functions with different values of shape parameter n,
we show that the width of the quality loss function depends upon the value
of n. In order words, if the value of n is increasing, then the slope of the
expected quality loss function is increasing. When we calculate the expected
quality loss, we consider the normal probability distribution. Similar results
are obtained with the exponential distribution, truncated exponential distri
bution, and truncated normal distribution. To show applicability of the pro
posed general quality loss function to the periodicity of upgrading fielded
systems, we present the quality loss function and demonstrate a process for
determining acceptance level of periodicity through numerical examples.
Therefore, the proposed general quality loss function can be used to justify
a decision to release a product upgrade, and determine a specification limit
on the release dates that minimizes the expected quality loss.
The limitations of this study are as follows: (1) We adopt Taylor series expan
sion for the general quality loss function. Due to this, the expected quality loss
using the proposed function has nominal errors. (2) Since we have difficulty in
obtaining actual data for quality loss associated with upgrading products and
periodicity, we cannot present a validation of shape parameter value, n.
References
Arts, G. R. J. 1998. Test Limits in Quality Control Using Correlated Products Characteristics,
ISBN 90-365-1129-1, The Netherlands: University of Twente.
Boehm, B. and Basili, V. R. 2001. Software defect reduction top 10 list. Software
Management Journal January: 135–137.
Chaplain, C. T. 2008. GAO -08-581T 24, U.S. Government Accountability Office,
April.
Choi, D. O. and Langford, G. 2008. A General Quality Loss Function Development
and Its Application to the Acquisition Phases of the Weapon Systems,
NPS-SE-08-007, Technical Report, Naval Postgraduate School, November.
Hutter, M. 2001. General Loss Bounds for Universal Sequence Prediction, Technical
Report IDSIA-03-01, Instituto Dalle di Studi sull’Intelligenza Artificiale, MannoLugano, Switzerland, April.
Conclusion
A general quality loss function having a shape parameter is developed
which is applicable to evaluate the expected quality loss for quality charac
teristics such as nominal-the-best, smaller-the-better, and larger-the-better.
Additionally, we present an appropriate range of shape parameter values in
the proposed general quality loss function to accommodate the impacts of
upgrades to fielded systems.
By plotting the loss functions with different values of shape parameter n,
we show that the width of the quality loss function depends upon the value
of n. In order words, if the value of n is increasing, then the slope of the
expected quality loss function is increasing. When we calculate the expected
quality loss, we consider the normal probability distribution. Similar results
are obtained with the exponential distribution, truncated exponential distri
bution, and truncated normal distribution. To show applicability of the pro
posed general quality loss function to the periodicity of upgrading fielded
systems, we present the quality loss function and demonstrate a process for
determining acceptance level of periodicity through numerical examples.
Therefore, the proposed general quality loss function can be used to justify
a decision to release a product upgrade, and determine a specification limit
on the release dates that minimizes the expected quality loss.
The limitations of this study are as follows: (1) We adopt Taylor series expan
sion for the general quality loss function. Due to this, the expected quality loss
using the proposed function has nominal errors. (2) Since we have difficulty in
obtaining actual data for quality loss associated with upgrading products and
periodicity, we cannot present a validation of shape parameter value, n.
References
Arts, G. R. J. 1998. Test Limits in Quality Control Using Correlated Products Characteristics,
ISBN 90-365-1129-1, The Netherlands: University of Twente.
Boehm, B. and Basili, V. R. 2001. Software defect reduction top 10 list. Software
Management Journal January: 135–137.
Chaplain, C. T. 2008. GAO -08-581T 24, U.S. Government Accountability Office,
April.
Choi, D. O. and Langford, G. 2008. A General Quality Loss Function Development
and Its Application to the Acquisition Phases of the Weapon Systems,
NPS-SE-08-007, Technical Report, Naval Postgraduate School, November.
Hutter, M. 2001. General Loss Bounds for Universal Sequence Prediction, Technical
Report IDSIA-03-01, Instituto Dalle di Studi sull’Intelligenza Artificiale, MannoLugano, Switzerland, April.