D.S. Chulani, "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Ph.D. Qualifying Exam Report, USC, February, 1998.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....estimation techniques have been developed. These techniques may be grouped into two major categories [25] algorithmic models, and non algorithmic models. The first category holds the most popular technique (at least in the literature) and is illustrated by estimation models such as COCOMO [5, 6, 8], PUTNAM SLIM[23] and function points analysis [2, 20] Algorithmic models are derived from the statistical or numerical analysis of historical projects data (simple multiple stepwise regression, Bayesian approach, polynomial interpolation, etc) There are two main disadvantages to these models. ....

....are measured on an ordinal or nominal scale composed of linguistic values. For example, the well known COCOMO 81 model has 15 attributes out of 17 (22 out of 24 in the COCOMOII) which are measured with six linguistics values: very low , low , nominal , high , very high , and extra high [5, 6, 8]. To overcome this limitation, in the next section we present a new method that can be seen as a fuzzification of the classical analogy to deal with linguistic values. 4. Estimation by Fuzzy Analogy Fuzzy Analogy is a fuzzification of the classical analogy procedure [15] It is also composed of ....

D.S. Chulani, "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Ph.D. Qualifying Exam Report, USC, February, 1998.

.... composed of qualifications such as very low and low (linguistic values in fuzzy logic) For example, in the COCOMO 81 model, 15 attributes out of 17 (22 out of 24 in the COCOMO II model) are measured with six linguistic values: very low , low , nominal , high , very high and extra high [2,3,4]. Another example is the Function Points measurement method, in which the level of complexity for each item (input, output, inquiry, logical file or interface) is assigned using three qualifications ( low , average and high ) Then there are the General System Characteristics, the calculation ....

D.S. Chulani, "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Ph.D. Qualifying Exam Report, USC, February, 1998.

....new projects, it is trained by a set of combination of inputs and outputs that are known as the training data. Our experiment consists in estimating the software development effort by using the neural networks approach on the COCOMO 81 dataset. The COCOMO 81 dataset contains 63 software projects [2,3,4]. Each project is described by 17 attributes: the software size measured in KDSI (Kilo Delivered Sources Instructions) the project mode is defined as either organic , semi detached or embedded , and the remaining 15 attributes are measured on a scale composed of six linguistic values: very ....

D.S. Chulani, "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Ph.D. Qualifying Exam Report, USC, February, 1998.

....COCOMO II provides a fixed set of estimates (namely effort and duration) for a fixed set of development phases. The extension of the capabilities of COCOMO II requires to extend the model, as is being done with COQUALMO, the model that estimates the number of introduced and removed errors [13]. On the contrary, our approach considers, estimates and combines the contributions of every single development activity in a given process. This finer granularity represents a way to account and adjust intrinsically for idiosyncrasies found in a project and its process; hence it can capture ....

Sunita Chulani, "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Ph.D. Qualifying Exam Report n. 98056 University of Southern California, Feb. 1998.

.... such as very low and low (linguistic values in fuzzy logic) For example, in the COCOMO 81 model (respectively the COCOMO II model) 15 attributes among 17 (respectively 22 among 24) are measured with six linguistic values ( very low , low , nominal , high , very high , extrahigh ) [2,3,4]. As another example, in the size measurement method of Function Points, the assignment of the level of complexity for each item (input, output, inquiry, logical file, or interface) uses three qualifications ( simple , average , complex ) and the calculation of the General System Characteristics ....

Chulani D., S., `Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension', PhD. Qualifying Examen Report, USC, February, 1998.

....is in many ways still as much art as engineering. Such data gathering difficulties make developing software cost estimation models particularly challenging. To get around this problem, the researchers at the USC Center for Software Engineering have evolved a multi step modeling methodology [5, 6] that we have found very useful for developing software estimation models when the amount of related empirical data is initially minimal. Using Bayesian statistical techniques, this approach allows us to establish initial model parameter values based on a blending of numbers derived from expert ....

Chulani, S., "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and its Quality Model Extension," USC-CSE tech. report 98-506.

....refered to as informal or intuitive cost estimation. 3 In this particular article, an informal basis is defined as intuition, comparison to similar, past projects based on personal memory, and guessing. 8 opinion through the use of Bayesian statistics to improve their predictive performance (Devnani Chulani, 1997). Perhaps the strongest statement that we have found in support of subjective estimates of cost, or at least of their utility, was made by Hughes (Hughes, 1996) There, he chides the negative perception that subjective cost estimates have in the research community, and attempts to partially ....

Devnani-Chulani, S., 1997. Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension. Technical Report, University of Southern California, Computer Science Department.

....and sample information provided a workable initial model, we seek a formal framework which explicitly recognizes that information about individual effort multipliers and scale factors varies considerably. A Bayesian analysis with an informative prior provides such a framework. Bayesian analysis [4,5], a mode of inductive reasoning used in many scientific disciplines, permits the investigator to combine sample (data) and prior (expert judgement) information in a logically consistent manner. Using Bayes theorem, prior (or initial) values are transformed to postdata views of the model ....

Chulani S., "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Qualifying Exam Report, Computer Science Department, USC Center for Software Engineering, February 1998.

No context found.

D.S. Chulani, "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Ph.D. Qualifying Exam Report, USC, February, 1998.

No context found.

Chulani98 - "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Sunita Chulani, Qualifying Exam Report, Computer Science Department, USC Center for Software Engineering, February 1998.

No context found.

Chulani98 - "Incorporating Bayesian Analysis to Improve the Accuracy of COCOMO II and Its Quality Model Extension", Sunita Chulani, Qualifying Exam Report, Computer Science Department, USC Center for Software Engineering, February 1998.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC