This paper aims at comparing diverse data mining techniques (e.g., derived from machine learning) for developing effective software quality prediction models. To achieve this goal, we tackled various issues, such as the collection of soft- ware metrics from open source repositories, the assessment of prediction mod- els to detect software issues and the adoption of statistical methods to evaluate data mining techniques. The results of this study aspire to identify the data min- ing techniques that perform better amongst all the ones used in this paper for software quality prediction models.
programmers. Among such models, those estimating software effort have motivated considerable research in recent years . Correct prediction of the software quality or maintain a software system is one of the most critical activities in managing software project. Due to the nature of the software engineering domain, it is important that software quality estimation models should be able to deal with ambiguity and indistinctness associated with such values. To serve this purpose, we propose our case- based estimation model for software quality estimation. We feel that case-based models are particularly useful when it is difficult to define concrete rules about a problem domain in addition to this, expert advice may be used to supplement the existing stored knowledge. A case-based reasoning model was developed in  for estimating software development effort.
Consider the quick improvement of theitem headway application get ready in show days. Programming headway application holds a couple of absconds in introducing/executing programming things. They are cost and intense development programming progression in testing the aftereffect of the item. Usually a level of the data burrowing frameworks were made for recognize programming deformation desire from various data set applications from obvious data. One pass count is one of the frameworks for getting to organizations and diverse idiosyncrasies of the planning units logically programming application headway including the tricks of programming application like thing expense and testing thing. Programming quality and testing profitability are the essential contrivances in programming blemish figure. So in this paper we propose to make insightful portrayal count to decrease cost of the item testing change and cost estimation for programming application process. This method propose to make programming quality and testing efficiency in by building perceptive modules from code attributes present in released thing sets. In this framework, utilize data association fundamental burrowing events for finding support and sureness for each data thing present dynamically programming application headway with property portrayal. This approach is help to engineers recognize programming absconds and bolster wander organization in assigning testing methods with resources feasibly. Keywords:Software defect production, association rule mining, classification, Defect testing, cost and database.
Quality of a software system depends on not only its functional but also its non-functional attributes. The prediction and determination of software quality of a component based system (CBS) becomes all the more important as the comprising components should be reusable. For that they must be reliable as well as reusable. Since quality is not something which is easily quantifiable it becomes a tedious task for conventional statistical models to predict software quality. Fuzzy logic can act as a great asset in these cases, where entities are closely related to the real world. An artificial neural network when combined with fuzzy inference system provided an architecture which can be trained and hence, is capable of predicting values. The said system has been employed for the purpose of quality prediction. Based on various factors several approaches have been proposed for determining and predicting software quality. But none of them use the combination of factors proposed in this paper.
Abstract—With the increasing complexity and size of software system, the difficulty of managing software quality is growing rapidly. How to ensure software quality has always been the important issue that needs to be solved. This paper builds a software quality predicting model for solving this issue. In order to realize this, it establishes OOPN (Object-oriented Petri nets) model to describe software development process which are organized as waterfall order. The stages are divided into three parts: previous stages, current stage and following stages. For the previous stages, the defects number could obtained by respective review activities. For the current and following stages, the defects number could predicted by the software quality regression model of each stage, which is built by using multiple linear regression method based on history data. At last, simulating the whole OOPN model and the Rayleigh function could be achieved by using the defects number of each stage. The Rayleigh function could be used to predict the defects number of software product.
Idri, A., Abran, A. and Khoshgoftaar, T. Fuzzy, 2001. Analogy: a New Approach for Software Cost Estimation. International (7) X. Huang, J. Ren and L.F. Capretz. A Neuro-FuzzyTool for Software Estimation .Proceedings of the 20th IEEE International Conference on Software Maintenance, p. 520 2004. Workshop on Software Measurement (IWSM´01), Montréal, Québec, Canada, August 28-29.
To begin with, this research defines Software Quality Prediction System (SQPS) as a system composed of a Classification Algorithm (CA) and a Software Quality Measurement Model (SQMM). Machine Learning applications in software quality measurement are expanding as research intensifies in two directions, the first direction focuses on improving the performance of CAs while the other direction concentrates on improving SQMMs. Despite of the increasing attention in this area, some well-designed SQPSs showed considerable false predictions, which could be explained by faults in the design of the CA, the SQMM, or the SQPS as a whole. In this context, there is a debate on which CA is better for measuring software quality, as well as there is a debate on which SQMM to follow. To start with, the research studied an original dataset of 7311 software projects. Then, the research derived quality measurements from the ISO 9126 Quality Model and developed the SQMM accordingly. Next, the research compared statistical measures of performance of four CAs , using WEKA and SPSS. Our experimental results showed that ISO 9126 is general, but flexible enough to act as a SQMM. Despite of their convergent performance, our experiments showed that Multilayer Perceptron Network (MLPN) have more balanced predictions than Naïve Bayes does. Following a rarely researched approach, the SQPS predicted five levels of software quality instead of making a binary prediction, limited with defect or non-defect software.
11 Read more
When the expression “Software Quality” is used, we usually think in terms of an excellent software product that fulfills our expectations. These expectations are based on the intended use. Number of models has been proposed for evaluation of software quality based on various characteristics. In this paper quality of software product is defined in terms of basic components as constituent part of any program or software and proposed a software quality prediction model based on basic components. It has been justified with example that if any software quality model uses the tacit knowledge that will be better than any other model in terms of quality.
Abstract— Recent years have witnessed a demand for software development era. In this paper presents a method to estimate the expected number of faults in the software in the early phase of its development and is based on the factors that influence software quality in the programming environment. The paper includes analysis of the five predicting variables and the required response variable to predict the early software reliability based on the previous data available. To predict the number of faults in the software before testing, data sets from previous projects are used. To evaluate the predictive capability of the developed model the predicted faults are compared with the actual faults. In this paper identifying the key influencing parameters are Techno- complexity, Practitioner level, Creation Effort, Review effort, Urgency.
Software fault prediction has a critical role to play in the software engineering framework which is required by the software developers for the implementation of an efficient software. Analysis and prediction of software faults would be a more tedious task that is highlighted in this technical work. Classification is one among the popular techniques utilized for software defect prediction. Its primary task involves the categorization of modules, denoted by a set of software metrics, into two classes that include fault-prone (FP), or non-fault-prone (NFP). For a certain classification model, the classifier has to be trained in prior on the basis of the training data acquired through the mining performed of past software archives, like change logs in software configuration management, bug reports in bug tracking systems, and the developers’ e-mails.
Abstract— A software fault prediction is a proven technique in achieving high software reliability. Prediction of fault-prone modules provides one way to support software quality engineering through improved scheduling and project control. Quality of software is increasingly important and testing related issues are becoming crucial for software. This necessitates the need to develop a real-time assessment technique that classifies these dynamically generated systems as being faulty/fault-free. A variety of software fault predictions techniques have been proposed, but none has proven to be consistently accurate. These techniques include statistical method, machine learning methods, parametric models and mixed algorithms. Therefore, there is a need to find the best techniques for Quality prediction of the software systems by finding the fault proneness. In this study, the performance of the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) is evaluated for Java based Object Oriented Software system from NASA Metrics Data Program (MDP) data repository on the basis of fault proneness of the classes.
that effective software development under method analysis will be performed. Author defined the analysis under the software complexity and method analysis so that the software design model will be improved. Author defined the development process with the specification of cost model under the size and volume analysis so that the prediction to the software system will be done effectively. Author defined the structural complexity model under integral factor so that the development effort will be reduced. Author defined a size and complexity based model for development of software system under cost estimation. Author defined an object oriented program so that effective software development will be done. Author defined a predictive software analysis under quality modeling so that the software system analysis will be done. Author defined an effective software evaluation and measurement for software system under traditional metrics analysis. Author defined a metrics suit with coupling, cohesion so that the software cost estimation will be done effectively.
Quality Analysis: Assessing the quality of a set of primary studies is a challenging task. A quality analysis questionnaire is prepared as part of this systematic mapping to assess the relevance of studies taking part in this mapping. The questionnaire takes into consideration suggestions given in Reference.  A total of 18 questions, given in Table 1, together form the questionnaire and each question can be answered as “Agree” (1 point), “Neutral” (0.5 points) and “Disagree” (0 points). Hence, a study can have a maximum 18 points and minimum 0 points. The same review committee enforces the quality analysis questionnaire.
12 Read more
Zigmund Bluvband (2011) describes that there are two advanced analytical models for obtaining accurate results for software reliability prediction. The First model can inhibit some specific features of software testing process and it is based on well-known S shaped Ohba model. This advanced model is applicable only for non- rare bug testing. For the rare bug rate prediction other model is proposed which is based on introduction of the additional control parameter last suspended time. He describes the easy way to comply with the conference paper formatting requirements is to use this document as a template and simply type your text into it.
There are five challenges or causes of agile outsource software projects and each cause has further sub-causes. The first and important cause is “Lack of source code access”. Currently after completion of milestone or project, source code is handed over first time to outsourcer but during project development there is no way to view and review source code. Without source code access and control, outsourcer is completely blind and he has to rely on outsourcee feedback. There is no way to monitor and find out what is actually happening and how many resources are working on project? Either committed software professionals are working on project or not?
14 Read more
In other words, the faults may be quantified by a software tool that can analyze the deltas in code modules maintained by the configuration control system and measure those changes specifically attributable to failure reports. When a fault was initially inserted into a component is based on the ability of the revision control system to identify the version in which each line first appeared in the module. For faults whose repair involves removing or modifying a line, determination is straightforward when the fault was introduced into the module.
16 Read more
Mrinal Singh Rawat et. al.(2012), identified causative factors which in turn suggest the remedies to improve software quality and productivity. They showed how the various defect prediction models are implemented resulting in reduced magnitude of defects. They presented the use of various machine learning techniques for the software fault prediction problem. The unfussiness, ease in model calibration, user acceptance and prediction accuracy of these quality estimation techniques demonstrate its practical and applicative magnetism. These modeling systems can be used to achieve timely fault predictions for software components presently under development, providing valuable insights into their quality. The software quality assurance team can then utilize the predictions to use available resources for obtaining cost effective reliability enhancements.
Abstract. Software engineering activities in the Industry has come a long way with various improve- ments brought in various stages of the software development life cycle. The complexity of modern software, the commercial constraints and the expectation for high quality products demand the accurate fault prediction based on OO design metrics in the class level in the early stages of software development. The object oriented class metrics are used as quality predictors in the entire OO software development life cycle even when a highly iterative, incremental model or agile software process is employed. Recent research has shown some of the OO design metrics are useful for predicting fault-proneness of classes. In this paper the empirical validation of a set of metrics proposed by Chidamber and Kemerer is performed to assess their ability in predicting the software quality in terms of fault proneness and degradation. We have also proposed the design complexity of object-oriented software with Weighted Methods per Class metric (WMC-CK metric) expressed in terms of Shannon entropy, and error proneness.
10 Read more
Dubey et al. (2012) proposed a methodology for quantifying the usability of software using a fuzzy multi- criteria weighted average approach. Fuzzy logic helped to deal with the uncertainty and imprecision of the importance and rating of attributes on which usability depends. This approach was chosen due to the highly unpredictable nature of the attributes on which usability depends. Ioannis (2013) presented an integrated solution through which significant improvement may be achieved, based on Multiple Criteria Decision Aid (MCDA) methodology and the exploitation of packaged software evaluation expertise in the form of an intelligent system. The MCDA methodology consists of different methods categorized into three classes, viz; multiple attribute utility method, outranking method and interactive method. Ioannis (2013) only addressed generalized issues of software evaluation that can be applied to diverse end user domains but was not specified to web-based software which is the main interest of this research. Alexiei (2014) proposed an Intelligent Usability Evaluation (IUE) tool to automate the usability evaluation process. Two set of tools - those predicting the usage of websites such as Cognitive Walkthrough for the Web (CWW) and those making use of conformance to standards such as Usability Evaluation framework (USEFUL) were used. These tools evaluated the usability of a website by employing the heuristic evaluation technique which references the set of research-based usability guidelines. However, there was no integration of the intelligent tool directly within website development environment. In this way, usability violations could surface in real time as the website is being created. This non intelligent approach involves the use of real persons; like the focused group and cognitive walkthrough. Farooq (2013), proposed a set of guidelines which define a protocol to carry out comparative study of software testing techniques. Certain factors were considered necessary for comparison during such test. These include number of faults, fault rate, fault type, size (test case generated), coverage, time (i.e. execution time), software/program type, experience of subjects, reliability improvement. The fact that a single technique cannot give the require result informs the selection of a combination of appropriate testing techniques to mitigate targeted software development faults.
10 Read more
Software testing is the most expensive and time consuming issue in the process of software development. It usually requires about 50% of the whole project schedule. On the other hand, it has been proved by experience that the majority of a system’s faults exist in a small fraction of modules. Thus, efficient prediction models are helpful for software testing. Accurate estimates of defective modules may help software developers in terms of allocating the limited resources and thus, decreasing testing times [3, 4].