Accurate Effort Estimation Research Articles

SEffEst: Effort estimation in software projects using fuzzy logic and neural networks

Abstract Academia and practitioners confirm that software project effort prediction is crucial for an accurate software project management. However, software development effort estimation is uncertain by nature. Literature has developed methods to improve estimation correctness, using artificial intelligence techniques in many cases. Following this path, this paper presents SEffEst, a framework based on fuzzy logic and neural networks designed to increase effort estimation accuracy on software development projects. Trained using ISBSG data, SEffEst presents remarkable results in terms of prediction accuracy.

A hybrid method for increasing the accuracy of software development effort estimation

Since software development environments, methods and tools are changing rapidly, the importance of accurate estimations in software projects is increasing significantly. Inaccurate estimations can lead to unpleasant results in the software projects so that many projects are failed at the early stages of the project. During the recent years, numerous estimation methods have been proposed that most of which are based on statistical techniques. Among all existing methods, simplicity of analogy based method makes it so common in this field. Analogy methods usually present accurate estimations but if the level of non normality in the software project datasets is high or type of most project features is categorical, these methods are confronted with inaccurate estimation problem. In this paper, genetic algorithm has been used under a new framework to improve the performance of analogy methods. A large dataset has been employed to evaluate the performance of the proposed method and the results have been compared with the other estimation methods. The results showed that the proposed method outperformed the other methods considerably

Guidelines for Software Development Effort Estimation

An analysis of 32 accuracy factors yields a list of useful guidelines for improving software development effort estimation accuracy.

EXPLORING THE EFFORT OF GENERAL SOFTWARE PROJECT ACTIVITIES WITH DATA MINING

Software project effort estimation requires high accuracy, but accurate estimations are difficult to achieve. Increasingly, data mining is used to improve an organization's software process quality, e.g. the accuracy of effort estimations. Data is collected from projects, and data miners are used to discover beneficial knowledge. This paper reports a data mining experiment in which we examined 32 software projects to improve effort estimation. We examined three major categories of software project activities, and focused on the activities of the category which has got the least attention in research so far, the non-construction activities. The analysis is based on real software project data supplied by a large European software company. In our data mining experiment, we applied a range of machine learners. We found that the estimated total software project effort is a predictor in modeling and predicting the actual quality management effort of the project.

Improving area swept estimates from bottom trawl surveys

Estimation of area swept is a key component for standardizing catch per unit effort (CPUE) data from fishery independent bottom trawl surveys and survey trawl gear experiments. Given technological advances and the proliferation of data streams from net mensuration equipment and global positioning system (GPS), techniques for estimating survey effort can be improved. Here we investigate new analytical techniques for improving the accuracy and precision of survey effort estimation. Sources of error and bias associated with two of the components used to compute area swept as a measure of fishing effort, distance fished by the trawl and net spread, are systematically examined and their influence quantified using both simulated and survey data. New analytical methods, a cubic spline smoothing algorithm to smooth GPS and net spread data, a haversine great circle algorithm to calculate distance between smoothed GPS track points, and a sequential outlier rejection algorithm to diminish the influence of noise on mean net spread estimates are shown to reduce or even eliminate the influence of biased observations on area swept estimators.

Linear combination of multiple case-based reasoning with optimized weight for software effort estimation

Since software development has become an essential investment for many organizations recently, both the software industry and academic communities are more and more concerned about a reliable and accurate estimation of the software development effort. This study puts forward six widely used case-based reasoning (CBR) methods with optimized weights derived from the particle swarm optimization (PSO) method to estimate the software effort. Meanwhile, four combination methods are adopted to assemble the results of independent CBR methods. The experiments are carried out using two datasets of software projects from Desharnais dataset and Miyazaki dataset. Experimental results show that different CBR methods can get the best results in different parameters settings, and there is not a best method for the software effort estimation among the six different CBR methods. Currently, combination methods proposed in this study outperform independent methods, and the weighted mean combination (WMC) method can get the better result.

Software project effort assessment

AbstractSoftware project assessments and postmortem analyses can increase the success of future projects and forthcoming project phases. However, although assessments and analyses are well‐presented in the software engineering literature, they are short of descriptions to assess effort. This paper proposes a stepwise effort assessment method that strives to facilitate learning from past experiences and to increase effort estimation accuracy by exploring realized project effort. The proposed method can be used either to assess the realized project effort of a time span from project's lifecycle or as a part of the whole project postmortem analysis in producing effort information for the project iteration assessment, final report, and annual project portfolio report. In addition, the effort information increases the project team's knowledge on realized activity sets and their relationships, and the information can be utilized in estimation method and model calibration for both re‐estimations and future estimations. The method was evaluated with case projects, a sample supplied by Tieto Finland Oy. Copyright© 2010 John Wiley & Sons, Ltd.

Two novel effort estimation models based on quality metrics in web projects

Web development projects are certainly different from traditional software development projects and, hence, require differently tailored measures for accurate effort estimation. Effort estimation accuracy will affect the availability of resource allocation and task scheduling. In this paper, we investigate the suitability of a newly proposed quality metrics and models to estimate the effort and duration for small or medium-size Web development projects. It then describes a new size metrics, presents two novel methods (WEBMO+ and VPM+), based on WEB model (WEBMO) using Web objects instead of SLOC and Vector Prediction Model (VPM), to fast estimate the development effort of Web-based information systems. We also empirically validate the approach with a four projects study. The results indicate that the approach provides a mechanism to measure the size of software systems, classify software systems, and estimate development effort early in the software life cycle to within +/-20 percent across a range of application types. In contrast with other existing methods, WEBMO+ and VPM+ uses raw historical information about development capability and high granularity information about the system to be developed, in order to carry out such estimations. This method is simple and specially suited for small or medium-size Web based information systems.

A Novel Algorithmic Cost Estimation Model Based on Soft Computing Technique

Problem statement: Software development effort estimation is the process of predicting the most realistic use of effort required for developing software based on some parameters. It has always characterized one of the biggest challenges in Computer Science for the last decades. Because time and cost estimate at the early stages of the software development are the most difficult to obtain and they are often the least accurate. Traditional algorithmic techniques such as regression models, Software Life Cycle Management (SLIM), COCOMO II model and function points, require an estimation process in a long term. But, nowadays that is not acceptable for software developers and companies. Newer soft computing techniques to effort estimation based on non-algorithmic techniques such as Fuzzy Logic (FL) may offer an alternative for solving the problem. This work aims to propose a new fuzzy logic realistic model to achieve more accuracy in software effort estimation. The main objective of this research was to investigate the role of fuzzy logic technique in improving the effort estimation accuracy by characterizing inputs parameters using two-side Gaussian function which gave superior transition from one interval to another. Approach: The methodology adopted in this study was use of fuzzy logic approach rather than classical intervals in the COCOMO II. Using advantages of fuzzy logic such as fuzzy sets, inputs parameters can be specified by distribution of its possible values and these fuzzy sets were represented by membership functions. In this study to get a smoother transition in the membership function for input parameters, its associated linguistic values were represented by two-side Gaussian Membership Functions (2-D GMF) and rules. Results: After analyzing the results attained by means of applying COCOMO II and proposed model based on fuzzy logic to the NASA dataset and created an artificial dataset, it had been found that proposed model was performing better than ordinal COCOMO II and the achieved results were closer to the actual effort. The relative error for proposed model using two-side Gaussian membership functions is lower than that of the error obtained using ordinal COCOMO II. Conclusion: Based on the achieved results, it was concluded that, using soft computation approaches such as fuzzy logic and their advantages, good predication; adaption; understandability and the accuracy of software effort estimation can be improved and the estimation can be very close to the actual effort. This novelty model will lead researchers to focus on benefits of non-algorithmic models to overcome the estimation problems.

Evaluating software cost estimation models using particle swarm optimisation and fuzzy logic for NASA projects: a comparative study

Bidding for contracts depends mainly on estimated costs of a given project, which makes an accurate estimation of effort and time required very important with great impact on budget computation and project success. Inaccurate estimates are likely lead to one or all of the following negative outcomes: failure in making a profit, increased probability of incomplete project and delay of project delivery date. In this paper, we provide a comparison between models developed for software cost estimation using particle swarm optimisation (PSO) algorithm, fuzzy logic (FL), and well-known cost estimation models such as Halstead, Walston-Felix, Bailey-Basili and Doty models. The performance of the developed models is evaluated based on the mean magnitude of relative error (MMRE) for NASA software projects.

Improving the Accuracy of Effort Estimation through Fuzzy Set Representation of Size

Problem statement: The precision and reliability of the effort estimation is very important for the competitiveness of software companies. The uncertainty at the input level of the Constructive Cost Model (COCOMO) yields uncertainty at the output, which leads to gross estimation error in the effort estimation. Fuzzy logic-based cost estimation models are more appropriate when vague and imprecise information was to be accounted for and was used in this research to improve the effort estimation accuracy. This study proposed to extend the COCOMO by incorporating the concept of fuzziness into the measurements of size. The main objective of this research was to investigate the role of size in improving the effort estimation accuracy by characterizing the size of the project using trapezoidal function which gave superior transition from one interval to another. Approach: The methodology adopted in this study was use of fuzzy sets rather than classical intervals in the COCOMO. Using fuzzy sets, size of a software project can be specified by distribution of its possible values and these fuzzy sets were represented by membership functions. Though, Triangular membership functions (TAMF) was used in the literature to represent the size, but it was not appropriate to clear the vagueness in the project size. Therefore, to get a smoother transition in the membership function, the size of the project, its associated linguistic values were represented by trapezoidal shaped MF and rules. Results: After analyzing the results attained by means of applying COCOMO, triangular and trapezoidal MF models to the COCOMO dataset, it had been found that proposed model was performing better than ordinal COCOMO and trapezoidal function was performing better than triangular function, as it demonstrated a smoother transition in its intervals and the achieved results were closer to the actual effort. The relative error for COCOMO using trapezoidal function is lower than that of the error obtained using TAMF. Conclusion: From the experimental results, it was concluded that, by fuzzifying the project size using TPMF, the accuracy of effort estimation can be improved and the estimated effort can be very close to the actual effort.

Updating weight values for function point counting

While software development productivity has grown rapidly, the weight values assigned to count standard Function Point (FP) created at IBM twenty-five years ago have never been updated. This obsolescence raises critical questions about the validity of the weight values; it also creates other problems such as ambiguous classification, crisp boundary, as well as subjective and locally defined weight values. All of these challenges reveal the need to calibrate FP in order to reflect both the specific software application context and the trend of todays software development techniques more accurately. We have created a FP calibration model that incorporates the learning ability of neural networks as well as the capability of capturing human knowledge using fuzzy logic. The empirical validation using ISBSG Data Repository (release 8) shows an average improvement of 22% in the accuracy of software effort estimations with the new calibration.

Improved estimation of software project effort using multiple additive regression trees

Accurate estimation of software project effort is crucial for successful management and control of a software project. Recently, multiple additive regression trees (MART) has been proposed as a novel advance in data mining that extends and improves the classification and regression trees (CART) model using stochastic gradient boosting. This paper empirically evaluates the potential of MART as a novel software effort estimation model when compared with recently published models, in terms of accuracy. The comparison is based on a well-known and respected NASA software project dataset. The results indicate that improved estimation accuracy of software project effort has been achieved using MART when compared with linear regression, radial basis function neural networks, and support vector regression models.

Avoiding Irrelevant and Misleading Information When Estimating Development Effort

Software projects average about 30 percent accuracy in effort estimation.1 Expecting highly accurate effort estimates might be unrealistic because software development projects are inherently uncertain. Nevertheless, software professionals' tendency toward overly optimistic estimates and their high level of estimation inconsistency suggest potential for improving effort estimation processes.

Forecasting of software development work effort: Evidence on expert judgement and formal models

The review presented in this paper examines the evidence on the use of expert judgement, formal models, and a combination of these two approaches when estimating (forecasting) software development work effort. Sixteen relevant studies were identified and reviewed. The review found that the average accuracy of expert judgement-based effort estimates was higher than the average accuracy of the models in ten of the sixteen studies. Two indicators of higher accuracy of judgement-based effort estimates were estimation models not calibrated to the organization using the model, and important contextual information possessed by the experts not included in the formal estimation models. Four of the reviewed studies evaluated effort estimates based on a combination of expert judgement and models. The mean estimation accuracy of the combination-based methods was similar to the best of that of the other estimation methods.

An investigation of artificial neural networks based prediction systems in software project management

A critical issue in software project management is the accurate estimation of size, effort, resources, cost, and time spent in the development process. Underestimates may lead to time pressures that may compromise full functional development and the software testing process. Likewise, overestimates can result in noncompetitive budgets. In this paper, artificial neural network and stepwise regression based predictive models are investigated, aiming at offering alternative methods for those who do not believe in estimation models. The results presented in this paper compare the performance of both methods and indicate that these techniques are competitive with the APF, SLIM, and COCOMO methods.

The adjusted analogy-based software effort estimation based on similarity distances

Analogy-based estimation is a widely adopted problem solving method that has been evaluated and confirmed in software effort or cost estimation domains. The similarity measures between pairs of projects play a critical role in the analogy-based software effort estimation models. Such a model calculates a distance between the software project being estimated and each of the historical software projects, and then retrieves the most similar project for generating an effort estimate. Although there exist numerous analogy-based software effort estimation models in literature, little theoretical or experimental works have been reported on the method of deriving an effort estimate from the adjustment of the reused effort based on the similarity distance. The present paper investigates the effect on the improvement of estimation accuracy in analogy-based estimations when the genetic algorithm method is adopted to adjust reused effort based on the similarity distances between pairs of projects. The empirical results show that applying a suitable linear model to adjust the analogy-based estimations is a feasible approach to improving the accuracy of software effort estimates. It also demonstrates that the proposed model is comparable with those obtained when using other effort estimation methods.

Optimization of analogy weights by genetic algorithm for software effort estimation

A reliable and accurate estimate of software development effort has always been a challenge for both the software industry and academia. Analogy is a widely adopted problem solving technique that has been evaluated and confirmed in software effort or cost estimation domains. Similarity measures between pairs of effort drivers play a central role in analogy-based estimation models. However, hardly any research has addressed the issue of how to decide on suitable weighted similarity measures for software effort drivers. The present paper investigates the effect on estimation accuracy of the adoption of genetic algorithm (GA) to determine the appropriate weighted similarity measures of effort drivers in analogy-based software effort estimation models. Three weighted analogy methods, namely, the unequally weighted, the linearly weighted and the nonlinearly weighted methods are investigated in the present paper. We illustrate our approaches with data obtained from the International Software Benchmarking Standards Group (ISBSG) repository and the IBM DP services database. The experimental results show that applying GA to determine suitable weighted similarity measures of software effort drivers in analogy-based software effort estimation models is a feasible approach to improving the accuracy of software effort estimates. It also demonstrates that the nonlinearly weighted analogy method presents better estimate accuracy over the results obtained using the other methods.

Does the linear size adjustment to estimated effort improve web applications effort estimation accuracy?

Over the last 16 years, and particularly over the last 8 years, Analogy-based effort estimation has been used to estimate effort for software projects and in several studies has presented comparable estimation accuracy to, or better than, algorithmic methods. The Analogy technique is also potentially easier to understand and apply by both researchers and practitioners. These two factors suggest that this technique has great potential as an effort estimation technique to be used within Companies. However, there are still several challenges, in particular regarding the type of effort adaptation to use in order to obtain the highest prediction accuracy, that need further investigation. Therefore this paper compares several methods of Analogy-based effort estimation and investigates the use of adaptation rules as a contributing factor to better estimation accuracy. Two datasets are used in the analysis; results show that the best predictions are obtained for the dataset that first, presents a continuous cost function, translated as a strong linear relationship between size and effort, and second, is more intact in terms of outliers and collinearity. Only one of the two types of adaptation rules employed generated good predictions.

An Evaluation of an Electronic Logbook as a More Accurate Method of Estimating Spatial Patterns of Trawling Effort and Bycatch in the Gulf of Mexico Shrimp Fishery

Abstract Penaeid shrimp trawling effort in the Gulf of Mexico is not measured directly but is estimated from landings and interview data gathered by port agents of the National Marine Fisheries Service (NMFS). The total pounds landed from each trip are assigned to one or more of the 219 statistical area-depth zone cells for which a catch rate (pounds per nominal day fished or catch per unit effort (CPUE)) is also obtained. Each month, the total landings in pounds for each cell are divided by the CPUE of the cell to determine effort or days fished. These values are summed across all cells and months to provide an estimate of total effort. The accuracy of total effort estimates is directly proportional to the accuracy of the landings allocations and estimated CPUE values. It has been assumed that allocation errors are random and that directional bias is negligible. In a previous paper, we described an electronic logbook (ELB) that enables accurate measurement of the spatial patterns of fishing effort. Herei...