Optimizing design parameters of fuzzy model based COCOMO using genetic algorithms
- PDF / 669,395 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 101 Downloads / 165 Views
ORIGINAL RESEARCH
Optimizing design parameters of fuzzy model based COCOMO using genetic algorithms Sonia Chhabra1 • Harvir Singh2
Received: 19 August 2018 / Accepted: 3 July 2019 Bharati Vidyapeeth’s Institute of Computer Applications and Management 2019
Abstract Software development process is a series of planned activities undertaken to design a software product. The major concern in this process is estimation of cost and effort. Algorithmic as well as non algorithmic techniques are used to estimate cost and effort. Algorithmic techniques use mathematical equations; however, in case of imprecise information these techniques are overpowered by non algorithmic techniques. Intermediate COCOMO suffers from a problem of imprecise definition of cost drivers resulting in inaccurate estimations. Thus in the current research, implementation of non algorithmic modelling is carried out using soft computing techniques like fuzzy logic and genetic algorithms. The fuzzy approach is implemented to design a fuzzy model for each cost driver. The fuzzy model handles imprecise and ambiguous definition of input ranges of cost drivers. Selection of parameters characterising fuzzy sets in proposed fuzzy model is further optimized using genetic algorithms. The proposed model is tested on COCOMO NASA dataset and COCOMO NASA2 dataset using MATLAB. The improvement in performance of proposed optimized model is measured in terms of mean magnitude of relative error (MMRE) and Pred (25%). A significant improvement in %MMRE and Pred (25%) justifies the suitability of genetic algorithms for optimizing proposed fuzzy model.
& Sonia Chhabra [email protected] Harvir Singh [email protected] 1
Uttrakhand Technical University, Dehradun, Uttrakhand, India
2
IIMT University, Meerut, Uttar Pradesh, India
Keywords Fuzzy logic Genetic algorithms Genetic tuning Software cost estimation Soft computing techniques Abbreviations ACO Ant colony optimization COCOMO Constructive cost model DB Data base EAF Effort adjustment factor EM Effort multiplier FIS Fuzzy inference system GA Genetic algorithms KB Knowledge base RB Rule base MF Membership function MMRE Mean magnitude of relative error MRE Magnitude of relative error TS Takagi–Sugeno
1 Introduction Software development process consists of number of distinct activities, each being executed in a specific order. Each activity requires an estimation of cost and effort to be applied. At early stages of project development, accurate estimate of development cost is very difficult. There is no short cut path leading to accurate estimation. Over the period of time, researchers have devised and used several methods for estimation. In general, all these methods can be classified under two broad categories [1] being algorithmic models and non-algorithmic models. Each model has its own characteristics along with respective strengths and weaknesses.
123
Int. j. inf. tecnol.
Algorithmic cost modelling methods consider mathematical expressions for carrying estimation process. The mathematical expr
Data Loading...
