Data mining in software metrics databases

Software metrics are measures of software. Their primary use is to help us plan and predict software development. We can better control software quality and development effort if we can quantify the software. This is especially true in the early stages of development. Research has been done in the area of predicting software maintenance effort from software metrics. In general, software metrics can be classified into two categories: software product metrics and software process metrics. Software product metrics are measures of software products such as source code and design documents. Software process metrics are measures of software development process. For example, the size of the software is a measure of the software product itself, thus a product metrics. The effort required to design a software system may be influenced by how the software is designed, thus a process metrics. In this article, only software product metrics are addressed.

Current software quality estimation models often involve the use of supervised learning methods for building a software fault prediction models. In such models, dependent variable usually represents a software quality measurement indicating the quality of a module by risk-basked class membership, or the number of faults. Independent variables include various software metrics as McCabe, Error Count, Halstead, Line of Code, etc... In this paper we present the use of advanced tool for data mining called Multimethod on the case of building software fault prediction model. Multimethod combines different aspects of supervised learning methods in dynamical environment and therefore can improve accuracy of generated prediction model. We demonstrate the use Multimethod tool on the real data from the Metrics Data Project Data (MDP) Repository. Our preliminary empirical results show promising potentials of this approach in predicting software quality in a software measurement and quality dataset.

The U.S. Geological Survey's automated information system's software for processing, storing, and retrieving water data is developed and maintained by the National Water Information System Program of the Water Resources Division. As part of that responsibility, a new system is currently being designed to replace the systems that run on a mainframe computer located at headquarters in Reston, Vz., and on minicomputers located in Division offices across the Nation. The new system is to run on workstations and file servers to be installed in all of the Division offices. A software life cycle provides the framework for defining, developing, and maintaining the new software system and for helping to meet software development challenges. Soft*'are quality assurance is needed in building the new system to provide confidence that the software is engineered to meet users' needs and conforms to established technical requirements. Successful software quality assurance depends on the application of uniform terminology and effective communication among users, developers, and managers. Software quality control is a planned and systematic process that focuses on establishing software standards and procedures, reviews and inspections, testing, and software configuration management throughout the software life cycle. This report explains the background for software quality assurance and the implementation of software quality assurance and quality control in the National Water Information System Program using a generic workbench approach, and presents an example workbench for the requirements-analysis phase of the software life-cycle.

Software metrics provide a quantitative basis for planning and predicting software development processes. Therefore the quality of software can be controlled and improved easily. Quality in fact aids higher productivity, which has brought software metrics to the forefront. This research paper focuses on different views on software quality. Moreover, many metrics and models have been developed; promoted and utilized resulting in remarkable successes. This paper examines the realm of software engineering to see why software metrics are needed and also reviews their contribution to software quality and reliability. Results can be improved further as we acquire additional experience with variety of software metrics. These experiences can yield tremendous benefits and betterment in quality and reliability.

Quality model is regarded as a well-accepted approach for assessing, managing and improving software product quality. There are three categories of quality models for software products, i.e., definition model, assessment model, and prediction model. Quality assessment model (QAM) is a metric-based approach to assess the software quality. It is typically regarded as of high importance for its clear method on how to assess a system. However, the current state-of-the-art in QAM research is under limited investigation. To address this gap, the paper provides an organized and synthesized summary of the current QAMs. In detail, we conduct a systematic mapping study (SMS) for structuring the relevant articles. We obtain a total of 716 papers from the five databases, and 31 papers are selected as relevant studies at last. In summary, our work focuses on QAMs from the following aspects: software metrics, quality factors, aggregation methods, evaluation methods and tool support. According to the analysis results, our work discovers five needs that researchers in this area should continue to address: (1) new method and criteria to tailor a quality framework (i.e., structure of software metrics and quality factors) according to different specifics, (2) systematic investigations on the effectiveness, strength and weakness of different aggregation methods to guide the method selection in different context, (3) more investigations on evaluating QAMs in the context of industrial cases, (4) further investigations or real-world case studies on the QAMs related tools, and (5) building a public and diverse software benchmark which can be adopted in different application context.

The selection of software metrics for building software quality prediction models is a search‐based software engineering problem. An exhaustive search for such metrics is usually not feasible due to limited project resources, especially if the number of available metrics is large. Defect prediction models are necessary in aiding project managers for better utilizing valuable project resources for software quality improvement. The efficacy and usefulness of a fault‐proneness prediction model is only as good as the quality of the software measurement data. This study focuses on the problem of attribute selection in the context of software quality estimation. A comparative investigation is presented for evaluating our proposed hybrid attribute selection approach, in which feature ranking is first used to reduce the search space, followed by a feature subset selection. A total of seven different feature ranking techniques are evaluated, while four different feature subset selection approaches are considered. The models are trained using five commonly used classification algorithms. The case study is based on software metrics and defect data collected from multiple releases of a large real‐world software system. The results demonstrate that while some feature ranking techniques performed similarly, the automatic hybrid search algorithm performed the best among the feature subset selection methods. Moreover, performances of the defect prediction models either improved or remained unchanged when over 85were eliminated. Copyright © 2011 John Wiley & Sons, Ltd.

Object-oriented software metrics are important in software engineering field for measurement of the software quality, size estimation, complexity and effort. Object oriented software development is the most popular and best practice in software development. Object oriented software metrics can be used for measuring extend of object orientedness of a program. These quantities are important information indicators, that can be helpful for a software developer in take decisions for designing good software. It is proven that the software metrics provide quantitative information about the structure of the code, which can be used to determine the reliability of software. The design of the software is highly recommended to be object oriented in the software development environment. Software metrics specific to programs that have object oriented design can be used to measure the quality of the software like inheritance, assessing the effects of coupling, the complexity of the code, size aspects, etc, of the software developed in such an environment. Software measurement is a way for assessing the quality of software. Many attributes can define quality of a software, internal (size, complexity) as well as external (reliability)[14]. Often internal attributes are contributing factors to external attributes. Reliability as a term means the degree to which a component, product or system executes specified functions for a specific period of time under specified conditions. Software reliability is the software being trustworthy of the specifications. The aim of the work is to model the relationship between reliability of object oriented software to the object oriented design metrics and predict reliability of software.

Software metrics are used to measure different attributes of software. To practically measure software attributes using these metrics, metric thresholds are needed. Many researchers attempted to identify these thresholds based on personal experiences. However, the resulted experience-based thresholds cannot be generalized due to the variability in personal experiences and the subjectivity of opinions. The goal of this paper is to propose an automated clustering framework based on the expectation maximization (EM) algorithm where clusters are generated using a simplified 3-metric set (LOC, LCOM, and CBO). Given these clusters, different threshold levels for software metrics are systematically determined such that each threshold reflects a specific level of software quality. The proposed framework comprises two major steps: the clustering step where the software quality historical dataset is decomposed into a fixed set of clusters using the EM algorithm, and the threshold extraction step where thresholds, specific to each software metric in the resulting clusters, are estimated using statistical measures such as the mean (μ) and the standard deviation (σ) of each software metric in each cluster. The paper’s findings highlight the capability of EM-based clustering, using a minimum metric set, to group software quality datasets according to different quality levels.

Software quality is one of the important aspects of the reliable software. For this reason, software practitioners are interested in the methods that enable to predict defect with software metrics. Most studies focus on what kind of techniques would be fit for the prediction model; however, the research about what is the appropriate number of software metrics for building effective defect prediction model are relatively infrequent. In this paper, we propose a new approach for selecting a subset of software metrics by exploiting the correlation structure among them. We construct software metrics networks with two different relationship measures of MIC (Maximal Information Coefficient) and Pearson correlation coefficient. We then adjust the threshold for filtering out edges with weak correlation, and select representative features for each of the resulting connected components. We validate this subset selection approach with the Poisson regression model. Our results demonstrate that metrics selection using networks with relationship coefficient is efficient enough compared to model with the full set of features.

The Software Quality Manager (SQM) aims at the construction of a software metrics evaluation environment in which the whole measurement process is performed via an engineering perspective. All the data, information and knowledge required for engineers during the process such as evaluation models and decision supporting knowledge is formally represented and managed to enhance consistency and to reduce misunderstanding. Concurrent measurement and analysis tasks occurring during the process are coordinated in SQM, in order to make the results significant. After defining the software metrics briefly, we present objectives and the three stage structure of SQM in detail. SQM supports qualitative/quantitative measurement techniques and is equipped with knowledge bases for software engineering and application domains. It also utilizes domain knowledge of applications in the form of software components. Finally, we present our ongoing efforts to make the method practical, including construction of two knowledge bases, integration of evaluation techniques, unified calibration units for software metrics, qualitative behavior generation techniques and so on.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

This paper shows that existing software metric tools interpret and implement the definitions of object-oriented software metrics differently. This delivers tool-dependent metrics results and has even implications on the results of analyses based on these metrics results. In short, the metrics-based assessment of a software system and measures taken to improve its design differ considerably from tool to tool. To support our case, we conducted an experiment with a number of commercial and free metrics tools. We calculated metrics values using the same set of standard metrics for three software systems of different sizes. Measurements show that, for the same software system and metrics, the metrics values are tool depended. We also defined a (simple) software quality model for maintainability based on the metrics selected. It defines a ranking of the classes that are most critical wrt. maintainability. Measurements show that even the ranking of classes in a software system is metrics tool dependent.

Software metrics can be an effective measurement tool to assess the quality of software. In the literature, there are a lot of software metrics applicable to systems implemented in different paradigms like Objects Oriented Programming (OOP). To guide the use of these metrics in the evaluation of the quality of software systems, it is important to define their thresholds. The aim of this study is to investigate the effectiveness of the thresholds in the evaluation of the quality of object oriented software. To do that, we used a threshold catalog of 18 software metrics derived from 100 software systems to define detection strategies for five bad smells. They are: Large Class, Long Method, Data Class, Feature Envy and Refused Bequest. We investigate the effectiveness of the thresholds in detection analysis of 12 software systems using these strategies. The results obtained by the proposed strategies were compared with the results obtained by the tools JDeodorant and JSPiRIT, used to identify bad smells. This study shows that the metric thresholds were significantly effective in supporting the detection of bad smells.

Summary form only given. The primary objective of introducing metrics and measurement is to improve software quality, and thus the productivity of the organization. Implementation of a software quality measurement effort, however, is not without cost. Therefore, it is important to understand that introducing software metrics and measurement may result in increasing up-front costs. Some consideration should be given to performing a cost benefit analysis to help determine if expectations are realistic and if the predefined goals are feasible. The identification of desired outcomes/results and the selection of appropriate measurement criteria are key to the success of a software improvement program.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Software metrics is a necessary step for software reliability and quality, and software metrics of traditional procedure-oriented programming are fairly mature and have various methodologies and tools available for use. Object-oriented programming has recently became popular. However, traditional procedure-oriented software metrics are not appropriate for the development of object-oriented software. Some research into object-oriented software metrics has been proposed, but these articles focus on only one metric that measures a specific characteristic of the object-oriented software. The authors propose a new metric methodology, the data scope complexity, for object-oriented software based on the data scope of a program. The data scope complexity can show complexities of multiple features of object-oriented programming at the same time. They also quantify and compare object-oriented programming with procedure-oriented programming.

A Framework for Managing, Measuring, and Predicting Attributes of Software Development Products and Processes Reflecting the immense progress in the development and use of software metrics in the past decades, Software Metrics: A Rigorous and Practical Approach, Third Edition provides an up-to-date, accessible, and comprehensive introduction to software metrics. Like its popular predecessors, this third edition discusses important issues, explains essential concepts, and offers new approaches for tackling long-standing problems. New to the Third Edition This edition contains new material relevant to object-oriented design, design patterns, model-driven development, and agile development processes. It includes a new chapter on causal models and Bayesian networks and their application to software engineering. This edition also incorporates recent references to the latest software metrics activities, including research results, industrial case studies, and standards. Suitable for a Range of Readers With numerous examples and exercises, this book continues to serve a wide audience. It can be used as a textbook for a software metrics and quality assurance course or as a useful supplement in any software engineering course. Practitioners will appreciate the important results that have previously only appeared in research-oriented publications. Researchers will welcome the material on new results as well as the extensive bibliography of measurement-related information. The book also gives software managers and developers practical guidelines for selecting metrics and planning their use in a measurement program.