Context: Software effort estimation is a very important component of the software development life cycle. It underpins activities such as planning, maintenance and bidding. Therefore, it has triggered much research over the past four decades, including many machine learning approaches. One popular approach, that has the benefit of accessible reasoning, is analogy-based estimation. Machine learning including analogy is known to significantly benefit from feature selection/weighting. Unfortunately feature weighting search is an NP hard problem, therefore computationally very demanding, if not intractable. Objective: Therefore, one objective of this research is to develop an effi cient and effective feature weighting algorithm for estimation by analogy. However, a major challenge for the effort estimation research community is that experimental results tend to be contradictory and also lack reliability. This has been paralleled by a recent awareness of how bias can impact research results. This is a contributory reason why software effort estimation is still an open problem. Consequently the second objective is to investigate research methods that might lead to more reliable results and focus on blinding methods to reduce researcher bias. Method: In order to build on the most promising feature weighting algorithms I conduct a systematic literature review. From this I develop a novel and e fficient feature weighting algorithm. This is experimentally evaluated, comparing three feature weighting approaches with a na ive benchmark using 2 industrial data sets. Using these experiments, I explore blind analysis as a technique to reduce bias. Results: The systematic literature review conducted identified 19 relevant primary studies. Results from the meta-analysis of selected studies using a one-sample sign test (p = 0.0003) shows a positive effect - to feature weighting in general compared with ordinary analogy-based estimation (ABE), that is, feature weighting is a worthwhile technique to improve ABE. Nevertheless the results remain imperfect so there is still much scope for improvement. My experience shows that blinding can be a relatively straightforward procedure. I also highlight various statistical analysis decisions which ought not be guided by the hunt for statistical significance and show that results can be inverted merely through a seemingly inconsequential statistical nicety. After analysing results from 483 software projects from two separate industrial data sets, I conclude that the proposed technique improves accuracy over the standard feature subset selection (FSS) and traditional case-based reasoning (CBR) when using pseudo time-series validation. Interestingly, there is no strong evidence for superior performance of the new technique when traditional validation techniques (jackknifing) are used but is more effi cient. Conclusion: There are two main findings: (i) Feature weighting techniques are promising for software effort estimation but they need to be tailored for target case for their potential to be adequately exploited. Despite the research findings showing that assuming weights differ in different parts of the instance space ('local' regions) may improve effort estimation results - majority of studies in software effort estimation (SEE) do not take this into consideration. This represents an improvement on other methods that do not take this into consideration. (ii) Whilst there are minor challenges and some limits to the degree of blinding possible, blind analysis is a very practical and an easy-to-implement method that supports more objective analysis of experimental results. Therefore I argue that blind analysis should be the norm for analysing software engineering experiments.

The term “Globalization” brought many challenges with itself in the field of software development. The challenge of accurate effort estimation in GSD is one among them. When talking about effort estimation, the discussion starts for effort estimation methods. There are a number of effort estimation methods available. Existing effort estimation methods used for co-located projects are might not enough capable to estimate effort for distributed projects. This is why; ratio of failure of GSD projects is high. It is important to calibrate existing methods or invent new with respect to GSD environment. This thesis is an attempt to explore the accuracy of effort estimation methods for distributed projects. For this purpose, the authors selected three estimation approaches: COCOMO II, SLIM and ISBSG. COCOMO II and SLIM are two well known effort estimation methods, whereas, ISBSG is used to check the trend of a project depending upon its (ISBSG’s) repository. The selection of the methods and approaches was based on their popularity and advantages over other methods/approaches. Two finished projects from two different organizations were selected and analyzed as case studies. The results indicated that effort estimation with COCOMO II deviated 15.97 % for project A and 9.71% for project B. Whereas, SLIM showed the deviation of 4.17% for project A and 10.86 % for project B. Thus, the authors concluded that both methods underestimated the effort in the studied cases. Furthermore, factors that might cause deviation are discussed and several solutions are recommended. Particularly, the authors state that existing effort estimation methods can be used for GSD projects but they need calibration by considering GSD factors to achieve accurate results. This calibration will help in process improvement of effort estimation.

Software maintenance is such an integral part of its evolutionary process that it consumes much of the total resources available. Some estimate the costs of maintenance to be up to 100 times the amount of developing a software. A software not maintained builds up technical debt, and not paying off that debt timely will eventually outweigh the value of the software, if no countermeasures are undertaken. A software must adapt to changes in its environment, or to new and changed requirements. It must further receive corrections for emerging faults and vulnerabilities. Constant maintenance can prepare a software for the accommodation of future changes. While there may be plenty of rationale for future changes, the reasons behind historical changes may not be accessible longer. Understanding change in software evolution provides valuable insights into, e.g., the quality of a project, or aspects of the underlying development process. These are worth exploiting, for, e.g., fault prediction, managing the composition of the development team, or for effort estimation models. The size of software is a metric often used in such models, yet it is not well-defined. In this thesis, we seek to establish a robust, versatile and computationally cheap metric, that quantifies the size of changes made during maintenance. We operationalize this new metric and exploit it for automated and efficient commit classification. Our results show that the density of a commit, that is, the ratio between its net- and gross-size, is a metric that can replace other, more expensive metrics in existing classification models. Models using this metric represent the current state of the art in automatic commit classification. The density provides a more fine-grained and detailed insight into the types of maintenance activities in a software project. Additional properties of commits, such as their relation or intermediate sojourn-times, have not been previously exploited for improved classification of changes. We reason about the potential of these, and suggest and implement dependent mixture- and Bayesian models that exploit joint conditional densities, models that each have their own trade-offs with regard to computational cost and complexity, and prediction accuracy. Such models can outperform well-established classifiers, such as Gradient Boosting Machines. All of our empirical evaluation comprise large datasets, software and experiments, all of which we have published alongside the results as open-access. We have reused, extended and created datasets, and released software packages for change detection and Bayesian models used for all of the studies conducted.

No description available.

Software effort estimation by analogy is a viable alternative method to other estimation techniques, and in many cases, researchers found it outperformed other estimation methods in terms of accuracy and practitioners' acceptance. However, the overall performance of analogy based estimation depends on two major factors: similarity measure and attribute selection & weighting. Current similarity measures such as nearest neighborhood techniques have been criticized that have some inadequacies related to attributes relevancy, noise and uncertainty in addition to the problem of using categorical attributes. This research focuses on improving the efficiency and flexibility of analogy-based estimation to overcome the abovementioned inadequacies. Particularly, this thesis proposes two new approaches to model and handle uncertainty in similarity measurement method and most importantly to reflect the structure of dataset on similarity measurement using Fuzzy modeling based Fuzzy C-means algorithm. The first proposed approach called Fuzzy Grey Relational Analysis method employs combined techniques of Fuzzy set theory and Grey Relational Analysis to improve local and global similarity measure and tolerate imprecision associated with using different data types (Continuous and Categorical). The second proposed approach presents the use of Fuzzy numbers and its concepts to develop a practical yet efficient approach to support analogy-based systems especially at early phase of software development. Specifically, we propose a new similarity measure and adaptation technique based on Fuzzy numbers. We also propose a new attribute subset selection algorithm and attribute weighting technique based on the hypothesis of analogy-based estimation that assumes projects that are similar in terms of attribute value are also similar in terms of effort values, using row-wise Kendall rank correlation between similarity matrix based project effort values and similarity matrix based project attribute values. A literature review of related software engineering studies revealed that the existing attribute selection techniques (such as brute-force, heuristic algorithms) are restricted to the choice of performance indicators such as (Mean of Magnitude Relative Error and Prediction Performance Indicator) and computationally far more intensive. The proposed algorithms provide sound statistical basis and justification for their procedures. The performance figures of the proposed approaches have been evaluated using real industrial datasets. Results and conclusions from a series of comparative studies with conventional estimation by analogy approach using the available datasets are presented. The studies were also carried out to statistically investigate the significant differences between predictions generated by our approaches and those generated by the most popular techniques such as: conventional analogy estimation, neural network and stepwise regression. The results and conclusions indicate that the two proposed approaches have potential to deliver comparable, if not better, accuracy than the compared techniques. The results also found that Grey Relational Analysis tolerates the uncertainty associated with using different data types. As well as the original contributions within the thesis, a number of directions for further research are presented. Most chapters in this thesis have been disseminated in international journals and highly refereed conference proceedings.

The estimation of effort is at the heart of project tasks, since it is used for many purposes such as cost estimation, budgeting, monitoring, project planning, control and software investments. Researchers analyze problems of the estimation, propose new models and use new techniques to improve accuracy. However up to now, there is no comprehensive estimation methodology to guide companies in their effort estimation tasks. Effort estimation problem is not only a computational but also a managerial problem. It requires estimation goals, execution steps, applied measurement methods and updating mechanisms to be properly defined. Besides project teams should have motivation and responsibilities to build a reliable database. If such methodology is not defined, common interpretation will not be constituted among software teams of the company, and variances in measurements and divergences in collected information prevents to collect sufficient historical information for building accurate models. This thesis proposes a methodology for organizations to manage and execute effort estimation processes. The approach is based on the reported best practices, v empirical results of previous studies and solutions to problems &amp / conflicts described in literature. Five integrated processes: Data Collection, Size Measurement, Data Analysis, Calibration, Effort Estimation processes are developed with their artifacts, procedures, checklists and templates. The validation and applicability of the methodology is checked in a middle-size software company. During the validation of methodology we also evaluated some concepts such as Functional Similarity (FS) and usage of Base Functional Components (BFC) in effort model on a reliable dataset. By this way we evaluated whether these subjects should be a part of methodology or not. Besides in this study it is the first time that the COSMIC has been used for Artificial Neural Network models.

Software effort estimation still remains a challenging and debatable research area. Most of the software effort estimation models take software size as the base input. Among the others, Constructive Cost Model (COCOMO II) is a widely known effort estimation model. It uses Source Lines of Code (SLOC) as the software size to estimate effort. However, many problems arise while using SLOC as a size measure due to its late availability in the software life cycle. Therefore, a lot of research has been going on to identify the nature of relationship between software functional size and effort since functional size can be measured very early when the functional user requirements are available. There are many other project related factors that were found to be affecting the effort estimation based on software size. Application Type, Programming Language, Development Type are some of them. This thesis aims to investigate the nature of relationship between software size and development effort. It explains known effort estimation models and gives an understanding about the Function Point and Functional Size Measurement (FSM) method. Factors, affecting relationship between software size and development effort, are also identified. In the end, an effort estimation model is developed after statistical analyses. We present the results of an empirical study which we conducted to investigate the significance of different project related factors on the relationship between functional size and effort. We used the projects data in the International Software Benchmarking Standards Group (ISBSG) dataset. We selected the projects which were measured by utilizing the Common Software Measurement International Consortium (COSMIC) Function Points. For statistical analyses, we performed step wise Analysis of Variance (ANOVA) and Analysis of Co-Variance (ANCOVA) techniques to build the multi variable models. We also performed Multiple Regression Analysis to formalize the relation. / Software effort estimation still remains a challenging and debatable research area. Most of the software effort estimation models take software size as the base input. Among the others, Constructive Cost Model (COCOMO II) is a widely known effort estimation model. It uses Source Lines of Code (SLOC) as the software size to estimate effort. However, many problems arise while using SLOC as a size measure due to its late availability in the software life cycle. Therefore, a lot of research has been going on to identify the nature of relationship between software functional size and effort since functional size can be measured very early when the functional user requirements are available. There are many other project related factors that were found to be affecting the effort estimation based on software size. Application Type, Programming Language, Development Type are some of them. This thesis aims to investigate the nature of relationship between software size and development effort. It explains known effort estimation models and gives an understanding about the Function Point and Functional Size Measurement (FSM) method. Factors, affecting relationship between software size and development effort, are also identified. In the end, an effort estimation model is developed after statistical analyses. We present the results of an empirical study which we conducted to investigate the significance of different project related factors on the relationship between functional size and effort. We used the projects data in the International Software Benchmarking Standards Group (ISBSG) dataset. We selected the projects which were measured by utilizing the Common Software Measurement International Consortium (COSMIC) Function Points. For statistical analyses, we performed step wise Analysis of Variance (ANOVA) and Analysis of Co-Variance (ANCOVA) techniques to build the multi variable models. We also performed Multiple Regression Analysis to formalize the relation. / +46-(0)-739763245

Background: Effort estimation is one of the cornerstones in project management with the purpose of creating efficient planning and the ability to keep budgets. Despite the extensive research done within this area, one of the biggest and most complex problems in project management within software development is still considered to be the estimation process. Objectives: The main objectives of this thesis were threefold: i) firstly to define the characteristics for a large project, ii) secondly to identify factors causing inaccurate effort estimates and iii) lastly to understand how the identified factors impact the effort estimation process, all of this within the context of large-scale agile software development and from the perspective of a project team.Methods: To fulfill the purpose of this thesis, an exploratory case study was executed. The data collection consisted of archival research, questionnaire, and interviews. The data analysis was partly conducted using the statistical software toolStata.Results: The definition of a large project is from a project team’s perspective based on high complexity and a large scope of requirements. The following identified factors were identified to affect the estimation process in large projects: deficient requirements, changes in scope, complexity, impact in multiple areas, coordination, and required expertise, and the findings indicate that these are affecting estimation accuracy negatively. Conclusions: The conclusion of this study is that besides the identified factors affecting the estimation process there are many different aspects that can directly or indirectly contribute to inaccurate effort estimates, categorized as requirements, complexity, coordination, input and estimation process, management, and usage of estimates.

Background. Effort estimation is today a crucial part of software development planning. However, much of the earlier research has been focused on the general conditions of effort estimation. Little to no effort has been spent on solution verification (SV) of the projects. It is not surprising considering that SV becomes more relevant, the larger the project. To improve effort estimation, it is key to consider the uncertainties from the assumptions and conditions it relies on. Objectives. The main objective of this study is to identify differences and similarities between general effort estimation and effort estimation in SV in order to find potential improvements to software project planning of large projects. More specifically, this thesis aims to identify what and how activities and factors affect effort uncertainty and what theory and methods can be applied to increase the accuracy of effort estimation in SV. Methods. An exploratory case study was conducted to reach the objectives. It was designed accordingly to the triangulation method and consisted of unstructured interviews, a questionnaire, and archival research. The analysis followed a four steps procedure. First, it aimed to identify each SV activity’s contribution to effort and effort uncertainty. Secondly, identify and analyze which and how factors impact the identified activities. Third, investigate the factors that impact effort uncertainty. Fourth and last, an analysis of how the factors and sources of uncertainty could be used to improve software project planning. Results. The result shows that the activities could be divided into two different groups, based on their difference in contribution to effort and effort uncertainty. The two activities showing a higher uncertainty than effort were trouble report handling& troubleshooting, which is by far the most uncertainty-causing, and fault correction lead-time. The fault-related factors were both collectively and non-collectively found to be the most uncertainty-causing. Furthermore, it showed that the type of product and what type of objective the employee has influenced the cause of uncertainty. Conclusions. The SV process shifts from a proactive and structured way to a more reactive and unstructured way of working with the project life cycle. Moreover, size is not a cause of uncertainty of effort, but the differences in products create different causes. It was concluded that to most effectively address inaccuracy in effort estimation, one should address the activities that constitute a minority in effort but the majority of uncertainty. The most straightforward approach to increase the performance of effort estimation in SV would be to evaluate the inclusion of fault prediction and fault correction. Consequently, the implementation of uncertainty identification and prevention methods such as the six Ws framework and the bottom-up/top-down effort estimation practices. / Bakgrund. Ansträngningsuppskattning är idag en viktig del av planeringen avmjukvaruutveckling. Mycket av den tidigare forskningen har fokuserat på demgenerella förhållandena av ansträngningsuppskattning. Lite till ingen energi har lagts på lösningverifiering av projekten. Det är inte förvånande med tanke på lösningsverifiering (LV) blir mer relevant, desto större projekt. För att förbättra ansträngninguppskattningen så är det viktigt att ta hänsyn till dem osäkerheter som härstammar från dem antaganden och förhållanden som den vilar på. Syfte. Huvudmålet av studien är att identifiera likheter och skillnader mellan den generella teorin om ansträngninguppskattning gentemot ansträngninguppskattning inom LV i avsikt att identifiera potentiella förbättringar av mjukvaruutvecklings planering för större projekt. Mer specifikt, åsyftade studien till att identifiera vilka och hur aktiviteter och faktorer påverkar ansträngnings osäkerheter, samt vilken redan existerande teori och modeller som skulle kunna appliceras för att öka noggrannheten i ansträngninguppskattningen av lösningverifieringen. Metod. En utforskande fallstudie genomfördes för att uppfylla målen. Den designades i enlighet med trianguleringsmetoden och bestod av ostrukturerade intervjuer, ett frågeformulär, samt en arkivstudie. Analysen följde en procedur på fyra steg. Det första steget hade i avsikt att identifiera varje aktivitets, i LV proccessen, tillförande av ansträngning och osäkerhet. Det andra steget avsåg att identifiera och analysera vilka och hur faktorer påverkade dem identifierade aktiviteterna. Det tredje steget åsyftade att undersöka dem faktorer som påverkar ansträngningsosäkerheten. Och slutligen, det fjärde steget avsåg att analysera hur dem identifierade faktorerna och källorna till osäkerhet kan användas för att förbättra mjukvaruprojekts planering. Resultat. Resultatet visar att aktiviteterna kunde kategoriseras i två olika grupper baserat på differensen mellan ansträngningen och den relaterade osäkerheten. Dem två aktiviteter som visade en högre osäkerhet än ansträngning var felrapports hantering & felsökning, som visade sig orsaka mest osäkerhet, samt ledtid till följd av felkorrigering. De felrelaterade faktorerna var både självständigt och kollektivt dem som skapar mest osäkerhetsgrundande. Samtidigt visade det sig att typ av produkt och vilken typ av arbete influerade grunden till osäkerhet. Slutsatser. LV proccesen skiftar med projekt livscykeln, från en proaktiv och strukturerad process, till en mer reaktiv och ostrukturerad process. Storlek är inte en grund för osäkerhet av ansträngning, däremot skapar skillnader mellan produkterna olika osäkerhetsgrunder. För att på ett så effektivt sätt som möjligt adressera felaktigheter i ansträngningsuppskattningarna, bör fokus lika på dem aktiviteter som utgör en minoritet av ansträngning och samtidigt utgör en majoritet i osäkerhet. Den mest självklara tillvägagångssättet för att öka prestandan av anstängningsuppskatningarna av LV är att evaluera införandet av fel deterktering och fel korrektion i modelen. Följaktligen, att implementera osäkerhetsidentifications och förhindrande metoder, till exempel "the six Ws framework" och "bottom-up/top-down" ansträningningsuppskattnings metoderna.

Performing a migration of a legacy system can often be a daunting task. However, there often comes a time where maintaining a legacy system is not profitable. At such a time, estimating how much effort is required to perform a migration can be vital for the legacy system holders. There is a lack of research that shows the effort distribution for migrations of small legacy systems.The contribution of this thesis is an effort distribution for a framework for migrations called Small System Migration Framework (SSMF) and SSMF. The purpose of the thesis is to evaluate how the effort is distributed over different activities when migrating a small legacy system. The goal of the thesis is to help provide a basis for the estimation process during migrations. This was done by documenting how effort is distributed over different activities contained in SSMF.This thesis takes an abductive approach, combining an inductive approach used in the creation of a framework and a deductive approach to document how effort was distributed during the migration. A framework was created using the literature study and this framework was used to conduct a migration.The result of this thesis was an updated framework and a table presenting the effort distribution of the migration. The framework showed factors that were influential when migrating the system. The effort distribution presents how effort is distributed over activities and shows which activities during the migration required more effort.Finally the thesis concludes that effort is highly centered around the preparation phase of the migration. Understanding legacy systems can be a challenge, lacking documentation and issues brought by the lack of maintenance results in high effort during this phase. Allocating more resources for the preparation phase and having access to people with experience during the preparation phase would likely make for a smoother transition with less unidentified problems appearing. / Att utföra en migration av ett ”legacy” system kan ofta vara en skrämmande uppgift. Det kommer dock ofta en tidpunkt då det inte längre är lönsamt att underhålla ett legacy system. Vid en sådan tidpunkt kan estimering av hur mycket insats som krävs för att utföra en migrering vara vital för ägarna av legacy systemet. Det finns en avsaknad av forskning som visar hur insats är fördelad för migrationer av små system.Bidraget av denna avhandling är ett ramverk för migrationer kallat Small System Migration Framework (SSMF) och en insats fördelning for SSMF. Ändamålet för avhandlingen är att evaluera hur insats är fördelad över olika aktiviteter vid migrering av små ”legacy” system. Målet med avhandlingen är att hjälpa förse en bas för estimeringsprocessen under migrering. Detta gjordes genom att dokumentera hur insats var fördelad över olika aktiviter i SSMF.Denna avhandling använde sig av ett abduktiv tillvägagångsätt, en kombination av ett induktivt tillvägagångssätt i skapandet av ett ramverk och ett deduktivt tillvägagångsätt i dokumenteringen av hur insats var fördelad under migrationen. En litteratur studie gjordes för att skapa ramverket och detta ramverk användes sedan för att göra en migrering.Resultatet av fallstudien var ett uppdaterat ramverk och en tabell som presenterar insatsfördelningen för en migrering. Ramverket visade faktorer som var inflytelserika vid migrering av systemet. Insatsfördelningen presenterade hur insats var fördelat mellan olika aktiviter och vilka aktiviteter som krävde mer insats under migreringen.Slutligen sammanfattar avhandlingen att insats är högt centrerad runt förberedelsefasen vid migrering. Att förstå legacy system kan vara en utmaning, bristande dokumentation och problem från bristande underhåll resulterar i hög insatsfördelning i denna fas. Allokering av mer resurser vid förberedelsefasen och att ha tillgång till personer med erfarenheter vid förberedelsefasen skulle troligen ge en mjukare övergång med mindre oidentifierade problem som visar sig.