International Journal of Engineering Trends and Technology (IJETT) - Volume4 Issue6- June 2013 Intelligent Models Creation for cost estimation using Clustering techniques Prof. N. Suresh HOD/CSE Kurinji College of Engineering and Technology , Manapparai – 621 307. Dr. D. Manimegalai M. E., Ph. D., Prof & Head / IT National Engineering College, Kovilpatti. Abstract: Software has played an increasingly important role in systems acquisition, engineering and development particularly for large complex systems. Software development may include research, new development, modification, reuse, re-engineering, maintenance or any other activities that result in software products. The software development methodology is a framework that is used to structure; plan and control the process of developing information systems. A wide variety of such frameworks have evolved over the years, each with its own recognized strengths and weakness. System development methodology is not necessarily suitable to be used by all projects, based on various technical, organizational, project and team considerations. Software development effort estimation is the process of predicting the most realistic use of effort required to develop or maintain software based on incomplete, uncertain and noisy input. Effort estimated may be used as input to project plans, iteration plans, budgets, and investment analyses, pricing processes and bidding rounds the term many ways of categorizing estimation, approaches. The top level categories are expert estimation, formal estimation, combinationbased estimation, statistical estimation. The ability to accurately estimate the time taken and cost for a project to come in to its successful conclusion is serious problem for software engineers. Accurate estimation of the software costs is a critical part of effective program management. The practice of predicting the cost of software has evolved, but it is far from perfect. The perfect cost-estimation method recommends an approach to improve the utility of the software cost estimated by exposing uncertainty (in understanding of the project as well as in costing accuracy) and reducing the risk that the estimate will be far different from the actual cost. My proposed work is to frame a cost estimation model in the approach to improve the utility of the software cost estimated by exposing uncertainty. I. Introduction: Software cost estimation is the difficult task and important task for successful projects. Using software cost estimation technique predict the time and cost to build successful software development. These kinds of operations are performed by client or developer in implementation. These operations are performs based on different parameters. Those parameters are project size and lines of code. Many software cost estimation techniques are introduced for development of ISSN: 2231-5381 high quality project. Existing software cost estimation techniques are predicts the less number of parameters. Those parameters are not gives the proper implementation steps environment. Previous Agile, COCOMO, COCOMO II are predicts the different kinds estimation techniques based on variables. Algorithm based estimation techniques are not gives the proper solution in implementation. After some number of days Non Algorithmic cost estimation is gives the proper solution in implementation specification. Now we are introduces in this paper generational http://www.ijettjournal.org Page 2350 International Journal of Engineering Trends and Technology (IJETT) - Volume4 Issue6- June 2013 evolutionary process. Its provide the single time cost estimation results. Refinement cost estimation procedure everything its possible using portfolio techniques. The above techniques are gives the better evidence based cost estimation techniques of content. Here we are show comparison results in between previous and present approaches in implementation. II. Related Work: Software effort estimation literature appeared from last three to four decades. All Successful projects are depends on effort and schedule. Previously different computational intelligence techniques are implemented for software effort estimation. Those techniques are fuzzy logic, neural networks and genetic programming. The above techniques are provides efficient results in software estimation. Sometimes these techniques provide the results are inaccurate efforts. Fig1: Effort Estimation Planning Firstly using dependent variables estimate the cost and effort values specification. These variables cost value is not sufficient apply the adjusting factors. All variables values are substitute and provide the efficient effort values. Adjusting factors provides re-scalable solution. Many times we are applying the adjusting ISSN: 2231-5381 factors and give the modified effort results. When we are provide the accurate solution no expert is not contains the idea. It is not provide optimal solution. Next apply the Function point. Its works based on project size. In project size start the estimation of cost procedure. It is not provide the satisfiable and accurate solution. Next in early stages based on lines of code (LOC) performs software cost estimation models. This approach also is not providing the optimal solution. Using FP and LOC start the estimation of cost models. These models are not gives the proper solution. Some other approaches are implemented for software cost estimation like algorithmic and non algorithmic environment. Algorithmic environment is manual rules estimation procedure. Sometimes it may take the wrong decisions. Compare to algorithmic, non algorithmic environments provides the best and efficient solution. Non Algorithmic environment itself applies the experienced and expert solution. In this present environment sometimes it may chance to apply the best opinions. After some number of days apply the expert judgment methods for software cost estimation. All different kinds of issues it’s possible to control using expert opinions. Software cost estimation works based on some data mining techniques. Data mining techniques also its possible estimate cost in efficient manner. Those data mining techniques are classification and decision trees. These data mining techniques give the solution under approximate software cost estimation process. Approximate cost analysis we are identifies based on rough set analysis. No such data mining techniques are not provides efficient cost estimation. In independent variables it’s not gives the proper estimation results. Now in new http://www.ijettjournal.org Page 2351 International Journal of Engineering Trends and Technology (IJETT) - Volume4 Issue6- June 2013 estimation techniques it applies the genetic algorithm. All variables are communication variables. It’s may possible to estimate the cost somewhat efficient in implementation process. Some kind of new investigation process it’s possible to relate to object oriented design. Using object oriented design it’s possible to estimate somewhat efficient. Object oriented design process uses as a reusable for identification of improved cost estimation. Neural networks and artificial intelligences are applied in software cost estimation process. In software cost estimation many number of layers are presented here. Verification of all the layers and provide the better results synchronization it’s very difficult and complex. Synchronization results are possible to identify based on learning algorithms. The above all techniques are not gives the proper cost estimation solution in implementation. It does not provide the perfect estimation information in software cost estimation process. Some other intelligence techniques are placed in implementation technique. Particle Swarm intelligence we are applied in implementation. Software Cost estimation approach applies iteratively and adding the new requirements. Its gives good software cost estimation techniques results. Negative software cost estimation converts into positive cost estimation process in implementation. These many numbers of times changes the requirements users are feel like complex or burden. III. Problem Statement: ISSN: 2231-5381 Previous many number of data mining techniques applied for software cost estimation. Many number of previous techniques are not update the software cost estimation parameters. Using limited number of software cost parameters provide the quality specification at most 50%. There is no new estimation parameters addition process in implementation process. Using exhaustive approach and refinement subset mechanism update the software cost estimation parameters. After updating of all efficient cost estimation parameters its provide the quality estimation results. All unnecessary estimation parameters are removed in implementation process. In each and every iteration apply the pruning techniques in implementation process. Group of group estimation parameters gives the nearest cost estimation optimal results. It is identifies from different analysis specifications mechanism. All groups of estimation parameters are quality estimation parameters. IV. Creation Intelligent Models for cost estimation using Clustering techniques: Many numbers of clustering techniques are in training projects environment. Using training projects identifies the new projects cost estimation is very easy. In total projects automatically categorize the number of projects based on training projects. Its gives the estimation results as a approximate manner specification process. Previously k-means clustering algorithm, k-mediods algorithms are implemented in software cost estimation. It’s not gives the proper cost estimation results. Next we are uses the Partitioning around mediods clustering techniques are preferred in software cost estimation. This type of clustering techniques applies exhaustively till gets the optimal solution. This estimation process applies http://www.ijettjournal.org Page 2352 International Journal of Engineering Trends and Technology (IJETT) - Volume4 Issue6- June 2013 hierarchically for showing the efficient results. In each and every step identifies the some kind of subset of results. Using from starting of the subset apply the refinement process. Perform the different iterations and gets the effectiveness subset results in output specification. and provide the better improved cost efficient effort estimation. In unsupervised mechanism learn the some new things and add into previous estimation approaches. These kinds of approaches are applied till quality cost effort estimation process. Using software cost estimation provides the quality project implementations. It’s not giving the quality learning results then we are apply the navies Bayesian technique. It’s is classify the estimation parameters are positive and negative preferences. In present software cost estimation mechanisms identifies the mean, variance and standard deviation values specification. Is there any values are available as a negative converts into positive environment with new cost effort estimation techniques. Every time changes the pattern based on project, it’s very complex for engineers in estimation environment. 4.2Pattern Approaches and Specific format Approaches: Fig 2: Clustering techniques based software cost effort estimation process Using K-means algorithm find out similar projects related to group, training projects are provides the good estimation parameters. K-means provides complete training results. Using those training results apply the testing approaches in implementation. 4.1Noisy Handling approaches for intelligent environment: In previous software cost effort estimation any errors are generate, its shows the noisy problems, those noisy results in effort environment implements with unsupervised clustering techniques. Using new unsupervised techniques adding the some rules for analyze ISSN: 2231-5381 Now we are introduces the new patterns for detection of software estimation mechanisms. Pattern based approach works for all kinds of projects in implementation process. Once we are define the software cost estimation parameters, no need to changes the software effort parameters. These pattern approaches give the optimization in implementation. It’s not providing the optimal solution then we are implement generational evolutionary algorithm. All software cost effort estimation parameters identifies based on rating approaches. In estimation approaches which kind of variables are uses in all estimation environment we are identifies. In total number of variables select one by one variable, and gives the rating. After utilization all software cost estimation models generate the one new generic http://www.ijettjournal.org Page 2353 International Journal of Engineering Trends and Technology (IJETT) - Volume4 Issue6- June 2013 model. Using this generic model start the estimation process. It is give proper solution alignment process. 6. At last it’s providing the best variables selection process related software cost estimation. 4.3Algorithm: Software cost effort estimation using generational evolution algorithm: In new generational evolutionary process applies the concept of portfolio management environment. Previous variables are not efficient, every time estimate the new variables also in implementation process. First here we are verifying different model of estimation variables. After all models variable analysis process find out the best variable selection. In how many models same variables are used find out based on similarity function. Using similarity function value specification finds out each and every variable fitness value. Using fitness values based variables select best variables. According to fitness values gives the specification of rating. These rating of variables are not sufficient automatically new variables also possible to evaluate and update the rating process. Its gives good evidence of software cost effort estimation results. This is completely related to generational evolutionary approach. The above all steps place into pseudo code. V.Performance Evolution: Fig 3: table values related k-means and k – generational evolutionary process 1. From different models start the analysis process and apply the similarity function – each and every variable of similarity variables results we are provide here. 2. This process applies till variables fitness values. This process applies till termination of all variables fitness values calculation. 3. After calculation of fitness provide the rating information. value 4. Rating variables start the allocation using reproduction operation. 5. Reproduction of variables is not gives the sufficient content then applies the replacement procedure. ISSN: 2231-5381 Fig 4: performance calculation in between of k-means and generational evolutionary process. VI.Conclusion: Compare present traditional clustering techniques to new generational evolutionary techniques are provides the best solutions in software cost effort estimation environment. Its can gives one kind good evidence. These things http://www.ijettjournal.org Page 2354 International Journal of Engineering Trends and Technology (IJETT) - Volume4 Issue6- June 2013 are efficient. Dynamic software cost estimation models we are implement using portfolio selection technique. Compare to previous techniques present techniques are works as a efficient techniques we are show the output. VII.References: 1. Ziauddin, Shahid Kamal Tipu, Shahrukh Zia, An Effort Estimation Model for Agile Software Development, ACSA, 2012 2. J.N.V.R.Swarup Kumar, A Novel Method for Software Effort Estimation Using Inverse Regression as firing Interval in fuzzy logic, IEEE.2011 3. Increasing the accuracy of software development effort estimation using projects clustering, IET, software, 2012 4. Arturo Chavoya, Applying Genetic Programming for Estimating Software Development Effort of Short-scale Projects, IEEE, 2011 5. Dirk Basten and Werner Mellis, A Current Assessment of Software Development Effort Estimation. IEEE, 2011 6. A.BalaKrishna , FUZZY AND SWARM INTELLIGENCE FOR SOFTWAREEFFORT ESTIMATION,AITM, 2012 7. Surendra Pal Singh ,A Review of Estimating Development Time and Efforts of Software Projects by Using Neural Network and Fuzzy, ijarcsse, 2012 8. Ziauddin ,An Effort Estimation Model for Agile Software Development, ACSA, 2012 9. Efi Papatheocharous, Computational Intelligence in Software Cost Estimation: Evolving Conditional Sets of Effort Value Ranges 10. Jin-Cherng Lin, Automatically Estimating Software Effort and Cost using Computing Intelligence Technique, CSCP,2012 ISSN: 2231-5381 http://www.ijettjournal.org Page 2355