A Simple Neural Network Approach to Software Cost Estimation

Software development effort estimation (SDEE) is one of the main tasks in software project management. It is crucial for a project manager to efficiently predict the effort or cost of a software project in a bidding process, since overestimation will lead to bidding loss and underestimation will cause the company to lose money. Several SDEE models exist; machine learning models, especially neural network models, are among the most prominent in the field. In this study, four different neural network models – Multilayer Perceptron, General Regression Neural Network, Radial Basis Function Neural Network, and Cascade Correlation Neural Network – are compared with each other based on: (1) predictive accuracy centered on the Mean Absolute Error criterion, (2) whether such a model tends to overestimate or underestimate, and (3) how each model classifies the importance of its inputs. Industrial datasets from the International Software Benchmarking Standards Group (ISBSG) are used to train and validate the four models. The main ISBSG dataset was filtered and then divided into five datasets based on the productivity value of each project. Results show that the four models tend to overestimate in 80% of the datasets, and the significance of the model inputs varies based on the selected model. Furthermore, the Cascade Correlation Neural Network outperforms the other three models in the majority of the datasets constructed on the Mean Absolute Residual criterion.

— The most challenging task of software project management is the cost estimation. Cost estimation is to accurately assess required assets and schedules for software improvement ventures and it includes a number of things under its wide umbrella, for example, estimation of the size of the software product to be produced, estimation of the effort required, and last but not the least estimating the cost of the project. The overall project life cycle is impacted by the accurate prediction of the software development cost. The COCOMO model makes employments of single layer feed forward neural system while being actualized and prepared to utilize the perceptron learning algorithm. To test and prepare the system the COCOMO dataset is actualized. This paper has the goal of creating the quantitative measure in not only the current model but also in our proposed model. Keywords— Software Cost Estimation, COCOMO, Artificial Neural Network, Feed Forward Neural Network, Magnitude Relative Error.

Software development effort estimation is the process of predicting the effort required to develop a software system. Estimating development effort accurately in the early stage of software life cycle plays a crucial role in effective project management. Effort estimation is a key factor for software project success, defined as delivering software of agreed quality and functionality within schedule and budget. Traditionally effort estimation has been used for planning and tracking project resources. It has become an important task. This paper proposed a neural network model for software effort estimation. This model has 3 layers. The train, validation and test data used are from COCOMO data set. Inputs and targets data randomly divided in train (60 %), validation (20%) and test (20%) group. When the number of neurons in hidden layer was 20, Number of training samples was 37, number of validation samples was 13 and number of testing samples was 13, the network has best performance. In this case, the value of training, validation and testing MSE was 0.01044, 0.0475 and 0.0375 respectively and value of training, validation and testing R was 0.9167, 0.7741 and 0.7410 respectively.

Journal of Cases on Information Technology

Software Engineering is a branch of computers that includes the development of structured software applications. Estimation is a significant measure of software engineering projects, and the skill to yield correct effort estimates influences vital economic processes, which include budgeting and bid tenders. But it is challenging to estimate at an initial stage of project development. Numerous conventional and machine learning-based methods are utilized for estimating effort and still, it is a challenge to achieve consistency in precise predictions. In this research exploration, various ANN-based models are compared with conventional algorithmic methods. The study also presents the comparison of results on various datasets from the artificial neural network models, deep learning models, higher-order Neural Network models, leading to the conclusion that hybrid methods yield better results. This paper also includes an analysis of primary data collected from Software Project professiona...

Procedia Computer Science, 2016

Software Effort Estimation models are hot topic of study over 3 decades. Several models have been developed in these decades. Providing accurate estimations of software is still very challenging. The major reason for such disappointments in projects are because of inaccurate software development norms; effort estimation is one such practice. Dynamically fluctuating environment of technology in software development industry make effort estimation further perplexing. One of the most commonly used algorithmic model for estimating effort in industry is COCOMO. Capability of machine learning particularly Artificial Neural Networks is to adjust a complex set of bond among the various independent and dependent variables. The paper proposes usage of ANN (Artificial Neural Network) based model technologically advanced using Multi Layered Feed Forward Neural Network which is given training with Back Propagation training method. COCOMO data-set is accustomed to test and train the network. Mean-Square-Error (MSE) and Mean Magnitude of Relative-Error (MMRE) are used as performance measurement indices. The experiment outputs suggest that the suggested model can provide better results and accurately forecast the software development effort.

Journal of Engineering and Technology, 2014

Last few decade software accomplishment admiration models developed, authentic estimates of the software activity beneath development is still unachievable goal. Recently advisers are alive on the development of new models and the advance of the absolute ones application bogus intelligence techniques. Designing of ANN (Artificial Neural Network) to archetypal a circuitous set of accord amid the abased capricious (effort) and the absolute variables (cost drivers) makes an apparatus for estimation. This cardboard presents an achievement assay of Multi ANNs in accomplishment estimation. We accept apish Back propagation ANN created by MATLAB Neural Network Apparatus application NASA dataset.

Software estimation is a process of estimating the effort, size and cost of the software, before the actual development of the software. Realistic estimation helps the software developers and managers in finding the required resources, man-hours, time and cost required in developing software. There are various types of estimation methods such as algorithmic, regression, machine learning, function pts etc, but still there is need of an accurate and precise software estimation method. Some models don't include all the parameters, some are extremely vulnerable to misclassification or error deviation is high. In this paper neural network training functions are used for software estimation as neural network is good at discovering relations (linear or non linear). Results of neural network are compared with cocomo 2 efforts as it is the most used model in the world for estimation. Input to neural network training functions is Cocomo 2 parameters which consist of 17 cost multipliers and 5 scale factors. After analyzing the results, a hybrid set of parameters are given to neural network training functions. Results from these hybrid parameters shows that neural networks techniques shows lowest mean error as compared to cocomo 2 parameters.

Software effort estimation is used to estimate how many resources and how many hours are required to develop a software project. The accurate and reliable prediction is the key to success of a project. There are numerous mechanisms in software effort estimation but accurate prediction is still a challenge for the researchers and software project managers. In this paper, the use of Neural Network techniques for Software Effort Estimation is discussed and evaluate on the basis of MMRE and Predicate. At the end, a specific Neural Network based Effort Estimation technique is proposed for further research.

Journal of Computer Science

Software development effort estimation becomes a very important and vital tool for many researchers in different fields. Software estimation used in controlling, organizing and achieving projects in the required time and cost to avoid the financial punishments due to the time delay and other different circumstances that may happen. Good project cost estimation will lead to project success and reduce the risk of project failure. In this paper, two neural network models are used, the Back-propagation algorithm versus the redial base algorithm. A comparison is done between the suggested models to find the best model that can reduce the project risks related to time and increase the profit by achieving the demands of the required project in time. The two models are implemented on a 60 of NASA public dataset, divided into 45 data samples for training and 15 data samples for testing. From the result obtained we can clearly say that the performance of the back-propagation neural network in training and testing cases is actually better than the radial base function, so the backpropagation algorithm can be recommended as a useful tool in the software effort and cost estimation.

2011

Failures of software are mainly due to the faulty project management practices, which includes effort estimation. Continuous changing scenarios of software development technology makes effort estimation more challenging. Ability of ANN(Artificial Neural Network) to model a complex set of relationship between the dependent variable (effort) and the independent variables (cost drivers) makes it as a potential tool for estimation. This paper presents a performance analysis of different ANNs in effort estimation. We have simulated four types of ANN created by MATLAB10 NNTool using NASA dataset.