Software projects are controlled by time, requirements and risks. Estimation of time and resources required for a project is a challenge due to the neglected effect or under/over quantification of intangible parameters like inexperience, unclear requirements, unfamiliar technologies, development environment, not bothering about the complexities in design and development. As a result the actual effort will be 2 to 3 times the order of magnitude of the estimated values. Rough estimates, which are 50 to 100% off the actual values, can be estimated when we are working with well-understood need and familiar with domain and technology issues. Fair estimates, which are 25 - 50% off the actual values, are possible to obtain when we know what needs to be done and have done many times before. Estimations made from Barry-Boehm, Bailey-Basili, COCOMO, Doty, Walston-Felix models gives different values for the same set of data because of the empirical relations differ. It is required to develop a method, which is suitable for the development environment and experience of the personal available in an organization.
Key words: Software Size, Risk, Effort, Estimation, Performance analysis, Model.
[...] We repeat the process of estimations using this effort unit matrix for the case studies presented in and Effort estimation for all the four cast studies is also made using the methods mentioned earlier and the results in person-months are tabulated in Table 4. TABLE 4 Actual & Estimated Effort in person-months Method KLOC Actual Effort AMN Yogi & Mala V Patil Bailey-Basili Barry-Boehm COCOMO II Doty model Walston Felix CS1 5.23 K CS2 1.50 K CS3 8.18 K CS K Where, symbols CS1, CS2 and CS3 stand for Case Study one, two and, three respectively, CS4 stands for the case study given in this paper. [...]
[...] TABLE 8a Quantification of Performance Probability Performance Drivers Complexity Size Stability Post Deployment Support Repair & Maintenanc e Average Requireme nts Performance Drivers Computer resources Personnel Standards Environment Constrai nts Performance Drivers Technology Performance Drivers Language Hardware Tools Data rights Experience Average Prototypes and reuse Documentati on Environment Management approach Integration Average Develo pment Approa ch Values of performance drivers related to Technology and development approach are again filled up based on the experience in Table 8b and it is observed that the average value for performance drivers relating to the technology component is 1.2 and development approach component is Therefore, quantified value for performance probability for all the four components of performance Requirements, Constraints, Technology and Development approach is Hence probability of risk is between 0.0 and Similarly the cost probability is worked and the values assigned to various cost drivers relating to the requirements, personnel are given in Table 9a and those for reusable software and the tools and environment are given in Table 9b. [...]
[...] On the other hand the historical information on past projects will definitely give us some idea, but the environment and technologies under which past projects were developed cannot compared with the present environment and technologies that are going to be utilized for the development of future projects SOFTWARE EFFORT ESTIMATION MODELS Numerous effort estimation models are developed based on size of the software. Size depends on either Estimated Kilo Lines Of Code (KLOC) or Function Points (FP). Among many KLOC-oriented estimation models the following are frequently referred in the literature Boehm simple model, Bailey-Basili model, COCOMO models, Doty model, Walston-Felix Model. [...]
[...] A method for effort estimation under 4GL environment was formulated by AMN Yogi and was applied by Mala V Patil for three different case studies [ for effort estimation. In this paper the formulated method is improved by appropriately changing the elements of effort unit matrix and is again applied to another case study. The results obtained for all the four case studies are compared with other estimation models. In addition, mean relative errors and standard deviations are estimated to compare the performance of these methods A CASE STUDY Project code-name “RYOJANA” was started after preliminary discussions with the concerned Users. [...]
[...] The performance of the present method based on the effort estimation process and other existing models is compared for the four case studies. Values obtained for Mean Relative and Root Mean Square Errors show that the methodology formulated by AMN Yogi and improved by Mala V Patil has the lowest MRE and RMSE values i.e and 9.88 respectively and hence its performance is considered as the best. It shows that the methodology formulated is able to provide good estimation capabilities. [...]
APA Style referenceFor your bibliography
Online readingwith our online reader
Content validatedby our reading committee