Formal estimation models There are many different types of formal effort estimation models available.
Effort estimation models may be based on sophisticated analyses and dependencies between effort and other variables in sets of previously completed projects and result in, e.g., formulae of the following type: Effort = a Sizeb * Adjustment factor
Examples of well-known estimation models are these:
COCOMO . In COCOMO, the use of effort of a typical project is assumed to follow a pre-defined formula regarding the relationship between Size and Effort. To account for differences in productivity, it is possible to make a number of (to some extent subjective) adjustments to the “nominal” effort estimate, e.g., the estimate can be adjusted for level of reuse.
Use Case Point Estimation (UCP) .The method takes as main input a software specification described through “Use Cases”, which is part of the notation of UML and is similar to the use of “Function Point”-based estimation models. Each use case is counted and weighted, and is adjusted for technical parameters. An expected productivity rate is provided as input. Current research in this area is typically directed at making better models, e.g., investigating when, how and how much local calibration of the models that are beneficial.
Model-based effort estimation processes may rely on expert judgment-based input. Hence, model output may also be biased towards over-optimism or be affected by the presence of irrelevant information. Further, inconsistency in effort estimates may be a problem, despite the use of models.
Effort Estimation Error Measurement:
Measuring estimation error is a fundamental activity in software estimation research. It is the basis of many activities, such as analysing whether or not an organization has an estimation problem, evaluating estimation methods, and identifying causes of estimation error. Software projects differ in size and, for aggregation purposes, most measures of estimation error are consequently based on relative estimation error. The most commonly used measure of estimation error is the Magnitude of Relative Error (MRE).
MRE is calculated by the following formula:
MRE= MOD(ACTUAL EFFORT-ESTIMATED EFFORT)/ACTUAL EFFORT.
REFERENCES:
1 .J. Li and G. Ruhe, "A comparative study of attribute weighting heuristics for effort estimation by analogy," proc. Proceedings of the 2006 ACM/IEEE international symposium on International symposium on empirical software engineering, pp. 66-74, 2006.
2. C. Lokan and E. Mendes, "Cross-company and Single-company Effort Models using the ISBSG Database: a Further Replicated Study," proc. Proceedings of the 2006 ACM/IEEE international symposium on International symposium on empirical software engineering, pp. 75-84, 2006.
3. S. Makridakis, S. C. Wheelwright, and R. J. Hyndman, Forecasting Methods and Applications, 3rd ed. New York: John Wiley & Sons, Inc., 1998.
Effort estimation models may be based on sophisticated analyses and dependencies between effort and other variables in sets of previously completed projects and result in, e.g., formulae of the following type: Effort = a Sizeb * Adjustment factor
Examples of well-known estimation models are these:
COCOMO . In COCOMO, the use of effort of a typical project is assumed to follow a pre-defined formula regarding the relationship between Size and Effort. To account for differences in productivity, it is possible to make a number of (to some extent subjective) adjustments to the “nominal” effort estimate, e.g., the estimate can be adjusted for level of reuse.
Use Case Point Estimation (UCP) .The method takes as main input a software specification described through “Use Cases”, which is part of the notation of UML and is similar to the use of “Function Point”-based estimation models. Each use case is counted and weighted, and is adjusted for technical parameters. An expected productivity rate is provided as input. Current research in this area is typically directed at making better models, e.g., investigating when, how and how much local calibration of the models that are beneficial.
Model-based effort estimation processes may rely on expert judgment-based input. Hence, model output may also be biased towards over-optimism or be affected by the presence of irrelevant information. Further, inconsistency in effort estimates may be a problem, despite the use of models.
Effort Estimation Error Measurement:
Measuring estimation error is a fundamental activity in software estimation research. It is the basis of many activities, such as analysing whether or not an organization has an estimation problem, evaluating estimation methods, and identifying causes of estimation error. Software projects differ in size and, for aggregation purposes, most measures of estimation error are consequently based on relative estimation error. The most commonly used measure of estimation error is the Magnitude of Relative Error (MRE).
MRE is calculated by the following formula:
MRE= MOD(ACTUAL EFFORT-ESTIMATED EFFORT)/ACTUAL EFFORT.
REFERENCES:
1 .J. Li and G. Ruhe, "A comparative study of attribute weighting heuristics for effort estimation by analogy," proc. Proceedings of the 2006 ACM/IEEE international symposium on International symposium on empirical software engineering, pp. 66-74, 2006.
2. C. Lokan and E. Mendes, "Cross-company and Single-company Effort Models using the ISBSG Database: a Further Replicated Study," proc. Proceedings of the 2006 ACM/IEEE international symposium on International symposium on empirical software engineering, pp. 75-84, 2006.
3. S. Makridakis, S. C. Wheelwright, and R. J. Hyndman, Forecasting Methods and Applications, 3rd ed. New York: John Wiley & Sons, Inc., 1998.
No comments:
Post a Comment