A Software Cost Estimation Taxonomy for Global Software Development

Projects

Manal El Bajta and Ali Idri

Software Project Management Research Team, ENSIAS, Mohammed V University of Rabat, Morocco

Keywords:

Software Cost Estimation, Global Software Development, Taxonomy.

Abstract:

Nowadays, software cost estimation plays an important role in the management and development of distributed

projects. The state of the art and cost estimation practice for Global Software Development (GSD) have

recently been identiﬁed. This knowledge has still not been structured. The objective of this paper is to structure

the knowledge about cost estimation for GSD. We used a design method to organize the knowledge identiﬁed

as a cost estimation taxonomy for GSD. The proposed taxonomy offers a classiﬁcation scheme for the cost

estimation of distributed projects. The cost estimation taxonomy consists of four dimensions: cost estimation

context, estimation technique, cost estimate and cost estimators. Each dimension in turn has multiple facets.

The taxonomy could then be used as a tool to developing a repository for cost estimation knowledge.

1 INTRODUCTION

Software development effort estimation for Global

Software Development (GSD) concerns the predic-

tion of the effort needed to develop a global software

project (Peixoto et al., 2010). Development effort

is considered to be one of the major components of

software costs, particularly as regards global develop-

ment, and is usually the most challenging to predict

(Lamersdorf et al., 2010), (Idri and Amazal, 2012).

To get a more comprehensive understanding of

how software cost estimation is practiced by dis-

tributed teams, a follow-up study is carried out to

elicit the state of the practice on cost estimation in

GSD projects (El Bajta et al., 2015), (El Bajta et al.,

2018). This study identiﬁed and aggregated knowl-

edge on cost estimation in GSD from the literature

and practice by means of a systematic review and

a survey respectively. The knowledge includes as-

pects such as approaches used to estimate cost in

GSD using size measures, cost drivers and the con-

text in which estimates are carried out. This body of

knowledge on cost estimation in GSD was structured

in order to launch both future research and improve

practice in this ﬁeld. A taxonomy is a classiﬁcation

scheme used in Software Engineering (SE) to orga-

nize the body of knowledge (Glass and Vessey, 1995).

The main purpose of this paper is to design a

software cost estimation taxonomy for GSD projects

that allows managers to accurately assess proposed

changes. Figure 1 illustrates the cost estimation pro-

cess to build the cost estimate taxonomy. It consists

on three steps: deﬁne cost drivers to be included in

the study, collect the extracted data and knowledge,

and build software cost estimation taxonomy.

This paper is structured as follows: we present re-

lated work in Section 2, identify the research method-

ology in Section 3, and report the results we obtained

in Section 4. We identify threats to validity in Section

5 and present conclusions and thoughts on the future

work in Section 6.

Figure 1: Cost estimation process.

2 RELATED WORK

Gumm (Gumm, 2006) developed a taxonomy of dis-

tribution to classify GSD projects dimensions. The

taxonomy was used to provide a basis for discussing

the challenges related to GSD projects and was based

on an earlier study performed by the same author

(Gumm, 2005). The proposed taxonomy uses four

different dimensions (physical distribution, organiza-

tional distribution, temporal distribution and distribu-

tion among stakeholder groups measured on a high-

medium- low scale) to classify the distribution of peo-

218

El Bajta, M. and Idri, A.

A Software Cost Estimation Taxonomy for Global Software Development Projects.

DOI: 10.5220/0007841202180225

In Proceedings of the 14th International Conference on Software Technologies (ICSOFT 2019), pages 218-225

ISBN: 978-989-758-379-7

ple and artifacts in GSD context.

Laurent et al. (Laurent et al., 2010) proposed a

taxonomy and a visual notation to address globally

distributed requirements engineering projects. The

main goal of the authors was to design the mod-

eling language, including site locations, stakeholder

roles, communication ﬂows, critical documents, and

supporting tools and repositories. The proposal was

based on the ﬁndings of eight in-depth interviews with

requirements analysts who worked on requirements

elicitation, analysis, and speciﬁcation tasks in glob-

ally distributed projects. These Interviews were per-

formed with the team leaders responsible for eliciting

and gathering the requirements in each project.

Smite et al. (

Smite et al., 2010) carried out a

study speciﬁcally focused on evidence of empirical

global software engineering. The study proposed a

classiﬁcation scheme to extract data from empirical

studies and systematize existing empirical global soft-

ware engineering studies. The proposed classiﬁcation

scheme helped to categorize the data extracted from

the study population, empirical background and re-

sults. The study revealed that the collection of papers

with empirical data on the subject of the desired sys-

tematic review was one of the main challenges, since

globally distributed project is at the cutting edge of

cross-disciplinary research. The process of deduct-

ing and collecting information about empirical back-

ground work was another difﬁculty reported. These

highlight the need for thorough descriptions of con-

texts in which empirical studies are conducted.

More recently, Smite et al. (

Smite et al., 2014)

have proposed a taxonomy of sourcing procedures.

The taxonomy proposed gives a typical terminol-

ogy and takes into consideration the classiﬁcation of

GSD projects with spotlight on the sourcing strate-

gies (e.g., Offshore outsourcing, offshore insourcing).

The result provided a systematically accumulated set

of terms categorized in the form of a taxonomy.

The taxonomy proposed by Smite et al. (

Smite

et al., 2014) is considered as knowledge classiﬁca-

tion approach. These taxonomy of sourcing strate-

gies is the most closely related work to ours, hence

our decision to use these taxonomy to understand the

GSD project setting. The base of these taxonomy is

more exhaustive, giving a more extensive scope of rel-

evant dimensions and clear criteria to classify GSD

projects. Furthermore, this taxonomy was also devel-

oped with the participation of several GSD experts,

which gives the taxonomy more credibility and va-

lidity. Therefore, our goal is to propose a taxonomy

to classify all dimensions of cost estimation for GSD,

and include categories related to empirical focus, sub-

jects of investigation and sources of data collection.

3 RESEARCH METHODOLOGY

This section describes the research question and

methodology used to design and evaluate the pro-

posed taxonomy.

3.1 Research Questions

This paper addresses one Research Question (RQ),

which is presented below:

RQ: How to organize the knowledge on cost esti-

mation for GSD projects?

The RQ is answered by organizing the cost esti-

mation knowledge for GSD projects as a taxonomy.

3.2 Taxonomy Design

The focus of this subsection is to present the method

to design the software cost estimation taxonomy used

in this study. Usman et al. (Usman et al., 2017)

present a revised and updated method on taxonomies

in the ﬁeld of software engineering. As shown in Ta-

ble 1, the method consists of four phases and thirteen

activities .

Phase 1: Planning

Planning represents the ﬁrst phase wherein basic deci-

sions about the taxonomy implementation and design

are made. In this phase, six activities are deﬁned as

shown in Table 1.

In activity A1, the SE knowledge area is selected

and described to make easier the understanding of the

context of the taxonomy and thus its application. The

taxonomy proposed in this paper is about cost estima-

tion for GSD context. During release and planning

phases, cost estimation plays an important role in the

management of distributed projects. Cost estimation

falls inside the extent of the ”Software engineering

management” knowledge area in SWEBOK version 3

(Bourque et al., 2014).

In activity A2, the main objectives and scope of

the taxonomy are to propose a classiﬁcation scheme

that can be used to describe software cost estimation

activities for GSD projects. A number of studies, in-

cluded in the Systematic Mapping Study (SMS) on

software cost estimation for GSD (El Bajta et al.,

2015) did not report signiﬁcant information on the

main context, techniques and also predictors used dur-

ing cost estimation. Researchers and practitioners

could therefore use the proposed taxonomy to con-

sistently report and recall important aspects related to

software cost estimation for GSD projects.

A Software Cost Estimation Taxonomy for Global Software Development Projects

219

Table 1: Taxonomy design method.

Phase Activity

Planning A1. Deﬁne SE knowledge area

A2. Describe the objectives of the taxonomy

A3. Describe the subject matter to be classiﬁed

A4. Select classiﬁcation structure type

A5. Select classiﬁcation procedure type

A6. Identify the sources of information

Identiﬁcation and extraction A7. Extract all terms

A8. Perform terminology control

Design and construction A9. Identify and describe taxonomy dimensions

A10. Identify and describe categories of each dimension

A11. Identify and describe the relationships

A12. Deﬁne the guidelines for using and updating the taxonomy

Validation A13. Validate the taxonomy

In activity A3, the subject matter for the classi-

ﬁcation deﬁnes what exactly is classiﬁed in the tax-

onomy. The cost estimation activities of the globally

developed projects is the subject of this taxonomy.

In activity A4, an appropriate classiﬁcation struc-

ture type is selected. Four basic classiﬁcation struc-

tures are deﬁned: hierarchy, tree, paradigm and

faceted analysis (Kwasnik, 1999). To structure our

taxonomy, faceted classiﬁcation is selected, since it is

suitable for evolving areas, such as software cost esti-

mation for GSD. The subject matter is classiﬁed from

several perspectives (facets) in faced classiﬁcation-

based taxonomies. Each facet of the proposed facets

is considered independent and has its own attributes,

making it easy to develop facet-based taxonomies

(Kwasnik, 1999).

In activity A5, an appropriate classiﬁcation pro-

cedure is determined. These type can be qualitative,

quantitative or both. Every facet of our taxonomy has

a set of values. The qualitative procedure is used to

select relevant facet values on the basis of extracted

data to characterize a speciﬁc estimation activity. In

some cases, it is impossible to assign a value simply

because of insufﬁcient data.

In activity A6, the data sources and data collec-

tion methods are identiﬁed to facilitate the prospec-

tion of knowledge related to the subject matter and

taxonomy. These data sources are selected from

peer-reviewed empirical studies on cost estimation for

GSD published in literature.

Phase 2: Identiﬁcation and Extraction

In this phase the relevant data required by the organi-

zation is identiﬁed and extracted. Two activities are

deﬁned as shown in Table 1.

In activity A7, the terms and concepts relevant to

the taxonomy are extracted from the sources identi-

ﬁed in the ﬁrst phase “planning”.

In activity A8, inconsistencies in the extracted

data are identiﬁed and removed.

Phase 3: Design and Construction

In this phase, the taxonomy is designed and con-

structed in order to identify dimensions and categories

in which the data items extracted can be organized.

Four activities are deﬁned as shown in Table 1.

In activity A9, the taxonomy dimensions are iden-

tiﬁed and described. They represent the main dimen-

sions or perspectives under which subject matter en-

tities are classiﬁed. Facet-based classiﬁcation taxon-

omy structure must have multiple dimensions (at least

two dimensions).

In activity A10, the categories for each of the di-

mensions are identiﬁed and described, each dimen-

sion must have at least two categories.

In activity A11, the relationships between cat-

egories and dimensions shall be identiﬁed and de-

scribed. Note that in some cases there are no rela-

tionship between dimensions, i.e. this activity could

be skipped.

In activity A12, the guidelines are provided to fa-

cilitate the adoption and evolution of the taxonomy.

Phase 4: Validation

Validation represents the last phase of this method to

ensure that the designed taxonomy is useful for users

to achieve their goals. In this phase, only one activity

is deﬁned as shown in Table 1.

In activity A13, the taxonomy can be validated

through benchmarking. Since there is no existing tax-

onomy on software cost estimation for GSD, bench-

marking our taxonomy against existing ones is impos-

sible.

ICSOFT 2019 - 14th International Conference on Software Technologies

220

4 RESULTS

In this section, we present the software cost esti-

mation taxonomy to answer the RQ: Organizing the

knowledge on cost estimation in GSD.

Taxonomies represent an effective tool to organize

and communicate the knowledge in an area (Glass

and Vessey, 1995). We have organized the identiﬁed

knowledge on cost estimation for GSD as a taxonomy.

The proposed taxonomy was developed according to

the method presented in Section 3. In this subsection,

we describe the results of this method, i.e. the soft-

ware cost estimation taxonomy.

Four dimensions are extracted and placed at the

top level of the taxonomy, as shown in Figure 2. A

dimension consists of all facets that are interrelated.

These four dimensions characterize the ﬁrst level of

taxonomy, and give an outline of the taxonomy at a

more extensive level.

Figure 2: First level of cost estimation taxonomy for GSD.

The estimation context represents the collection of

those facets that deﬁne and characterize the context in

which the estimation activity in a distributed project is

carried out. So as to fully characterize a speciﬁc soft-

ware cost estimation activity of GSD project, facets of

all dimensions should to be described. Each of these

four dimensions and their facets are described in de-

tail below.

4.1 Cost Estimation Context

The context is a central concept in empirical software

engineering. It is one of the distinctive features of

the discipline and it is an inseparable part of software

practice. Context refers to a broad perspective, and it

needs to be properly captured, reported , and contex-

tualized in the empirical SE studies to communicate

the applicability of the research ﬁndings. Thus, con-

text draws attention to what resources are nearby, and

when and where to use the reported ﬁndings (Dyb

et al., 2012), (Dey, 2001).

Seven facets are extracted from context dimen-

sion. These facets and their possible values are pre-

sented in Figure 3. They are described below:

• Planning: Estimation supports planning at vari-

ous levels in GSD context. This mainly includes

release and sprint planning (Hossain et al., 2009),

while some teams may also make estimates during

Figure 3: Cost estimation context dimension.

daily meetings. Project bidding is another level at

which companies must estimate the total develop-

ment cost in advance to bid for the projects.

• Project Activities: This facet shows which devel-

opment activities are accounted for in the estima-

tion of software costs. For example, The product

life cycle describes maintenance activity, or the

total cost estimate do not include the time spent

on maintenance.

• Project Domain: This facet captures the domain

of project for which the software cost develop-

ment is being estimated. Different domains could

lead to different sets of cost estimation. We have

used the categories from the project domains re-

viewed in previous study (El Bajta et al., 2017b).

• Project Setting: This facet represents the setting

in which the global software teams are develop-

ing the project. Smite et al. (

Smite et al., 2014)

proposed a global software engineering taxonomy

that characterizes two broad settings of the global

teams: onshore and offshore.

• Planning Approaches: This facet documents the

planning approach practiced by GSD team.

• Number of Sites: It records the number of sites

for the GSD project, and thus communicates im-

portant information related to the different sites

worldwide.

• Team Size: It documents the team size which is

responsible for developing the estimated tasks.

4.2 Estimation Technique

This dimension includes the facets that are related

to estimation techniques. Those facets should be re-

ported to characterize a GSD team’s estimation activ-

ity (El Bajta, 2015). Figure 4 describes the facets of

this dimension and their corresponding values.

A Software Cost Estimation Taxonomy for Global Software Development Projects

221

Figure 4: Estimation technique dimension.

• Estimation Technique: This documents the esti-

mation techniques applied for GSD projects. Ac-

cording to the SMS results (El Bajta et al., 2015),

the cost estimation techniques for GSD projects

are expert judgment, machine learning and non-

machine learning (Hosni et al., 2017), (El Bajta

et al., 2017a), (Amazal et al., 2014b), (Idri et al.,

2016d), (Idri et al., 2016c).

• Use Technique: In GSD, there are different types

of cost estimation techniques. An individual or a

group of experts can use these techniques. This

facet documents whether the effort was estimated

using an estimation technique based on individu-

als or groups (Amazal et al., 2014a), (Idri et al.,

2002b).

4.3 Cost Estimate

The main output of the estimation activity is cost es-

timates. The facets proposed in this dimension deﬁne

cost estimate. Those facets with their corresponding

values are presented in Figure 5.

Figure 5: Cost estimate dimension.

• Estimated Cost: This facet documents the esti-

mated cost that represents the main output of the

estimation activity.

• Actual Cost: It is important to have the actual cost

at the end of planning, to enable comparison with

the estimated cost.

• Estimation Dimension: This facet documents the

important and critical dimensions of estimation,

e.g. estimation of development effort as total ef-

fort hours.

• Accuracy Measure: This facet records cost perfor-

mance ways to assess the accuracy of the applied

estimation technique (Idri et al., 2016b).

4.4 Cost Estimators

Cost estimators play an important role in calculating

costs. They consist of cost drivers such as size, team

capabilities, product requirements, etc (Idri et al.,

2016a), (Idri and Abran, 2001). One of the most

required development cost is related to project size.

Five facets are collected regarding the cost estimators

dimension. These facets and their possible values are

presented in Figure 6.

Figure 6: Cost estimators dimension.

• Product Size: In general, the development cost

is strongly correlated with product size (Jaakkola,

2009). This facet documents whether distributed

teams use size as an estimator and which statisti-

cal analysis is used to represent this size.

• Team Experience: A development team experi-

ences with global software development projects

impact the required cost (Ramasubbu et al., 2011).

This facet describes whether a team experience

was considered or not in arriving at the cost es-

timates.

• Team Structure: Distribution of skills and team

structure impact the required effort (Agerfalk

et al., 2005). This facet documents whether the

structure of the team members was considered or

not during the cost estimation session.

• Product Requirement: Strict product require-

ments increase the development cost (Chung

et al., 2012), (Idri et al., 2002a). This facet records

ICSOFT 2019 - 14th International Conference on Software Technologies

222

which product requirements were considered in

arriving at cost estimates.

• Distributed Teams’ Distances: The geographi-

cal, temporal, and socio-cultural distances be-

tween global development teams increase the de-

velopment cost due to the increased complexity of

the collaboration and communication (Holmstrom

et al., 2006).

5 THREATS TO VALIDITY

The results of the proposed taxonomy may have been

inﬂuenced by the coverage of the study search, and

also the inaccuracy in study data extraction. Four

types of validity threats (Easterbrook et al., 2008) of

the study results are therefore discussed in the follow-

ing subsections.

Construct validity is concerned with issues caused

by poor data collection and recording, also, exacti-

tude of the interpretation of the concepts studied and

the completeness of the relevant studies collected. In

this study, the data identiﬁed and aggregated are ex-

tracted from the SMS (El Bajta et al., 2015) and sur-

vey (El Bajta et al., 2017b). The extracted data was

used as the main input for the taxonomy designed in

this study. To ensure the correct interpretation of these

data, we checked the contributions of the concepts in

related literature and all the authors discussed these

data in order to reach a consensus as to their use and

contribution.

Internal validity is concerned with analyzing ex-

tracted data. Threats to internal validity are impor-

tant as our study does not present an evaluation of the

taxonomy. In future studies, we plan to evaluate and

assess the taxonomy by using it to characterize esti-

mation cases from the literature and the industry.

Conclusion validity is to ensure that reasonable

conclusions are drawn on the basis of data collected

and that problems such as the bias of researchers do

not lead to incorrect conclusions. We used a tax-

onomy design method to systematically organize the

knowledge on software cost estimation for GSD as

taxonomy in a systematic manner.

External validity is concerned with the represen-

tativeness of the selected studies as regard the overall

goal of the study. The results of this study were con-

sidered with regard to the cost estimation for GSD

projects. These results can serve as a starting point

for researchers and practitioners working in this ﬁeld

to further improve the completeness and usefulness of

the proposed taxonomy.

6 CONCLUSIONS

The development of taxonomies helps to structure,

generalize and share existing knowledge and to ad-

vance research (Glass and Vessey, 1995). We have

organized the existing body of knowledge on cost es-

timation in GSD as a taxonomy. The taxonomy has

been systematically developed by following a taxon-

omy design method. One research question was ad-

dressed by incorporating ﬁve dimensions to organize

knowledge on cost estimation for GSD projects. The

main usage for our taxonomy is to provide a basis for

researchers to classify their own studies and related

studies on cost estimation for GSD ﬁeld. This taxon-

omy could therefore be used as a tool to develop a cost

estimation knowledge repository to better understand

and improve the cost estimation practice in the global

development context in the long term.

The usefulness of the taxonomy has not been

demonstrated in the study. We plan to apply the de-

veloped taxonomy on data extracted and reported in

the literature to characterize cost estimation cases.

REFERENCES

Agerfalk, P. J., Fitzgerald, B., Holmstrom Olsson, H.,

Lings, B., Lundell, B., and

O Conch

uir, E. (2005). A

framework for considering opportunities and threats

in distributed software development.

Amazal, F.-A., Idri, A., and Abran, A. (2014a). An analogy-

based approach to estimation of software development

effort using categorical data. In 2014 Joint Conference

of the International Workshop on Software Measure-

ment and the International Conference on Software

Process and Product Measurement, pages 252–262.

IEEE.

Amazal, F. A., Idri, A., and Abran, A. (2014b). Software de-

velopment effort estimation using classical and fuzzy

analogy: a cross-validation comparative study. Inter-

national Journal of Computational Intelligence and

Applications, 13(03):1450013.

Bourque, P., Fairley, R. E., et al. (2014). Guide to the soft-

ware engineering body of knowledge (SWEBOK (R)):

Version 3.0. IEEE Computer Society Press.

Chung, L., Nixon, B. A., Yu, E., and Mylopoulos, J. (2012).

Non-functional requirements in software engineering,

volume 5. Springer Science & Business Media.

Dey, A. K. (2001). Understanding and using context. Per-

sonal and ubiquitous computing, 5(1):4–7.

Dyb

a, T., Sjøberg, D. I., and Cruzes, D. S. (2012). What

works for whom, where, when, and why?: on the role

of context in empirical software engineering. In Pro-

ceedings of the ACM-IEEE international symposium

on Empirical software engineering and measurement,

pages 19–28. ACM.

A Software Cost Estimation Taxonomy for Global Software Development Projects

223

Easterbrook, S., Singer, J., Storey, M.-A., and Damian, D.

(2008). Selecting empirical methods for software en-

gineering research. In Guide to advanced empirical

software engineering, pages 285–311.

El Bajta, M. (2015). Analogy-based software development

effort estimation in global software development. In

IEEE 10th International Conference on Global Soft-

ware Engineering Workshops (ICGSEW), pages 51–

54. IEEE.

El Bajta, M., Idri, A., Fern

andez-Alem

an, J. L., Ros, J. N.,

and Toval, A. (2015). Software cost estimation for

global software development a systematic map and

review study. In Evaluation of Novel Approaches to

Software Engineering (ENASE), 2015 International

Conference on, pages 197–206. IEEE.

El Bajta, M., Idri, A., Ros, J. N., Fern

andez-Alem

an, J. L.,

De Gea, J. M. C., Garcıa, F., and Toval, A. (2018).

Software project management approaches for global

software development: a systematic mapping study.

Tsinghua Science and Technology, 23(6):690–714.

El Bajta, M., Idri, A., Ros, J. N., Fernandez-Aleman, J. L.,

and Toval, A. (2017a). Estimation of costs and time

for the development of distributed software. In Soft-

ware Project Management for Distributed Computing,

pages 25–42. Springer.

El Bajta, M., Idri, A., Ros, J. N., Fern

andez-Alem

an, J. L.,

and Toval, A. (2017b). Software cost attributes in

global software development projects: A survey. In

Proceedings of the 9th International Conference on

Information Management and Engineering, pages 96–

101. ACM.

Glass, R. L. and Vessey, I. (1995). Contemporary

application-domain taxonomies. IEEE Software,

(4):63–76.

Gumm, D. C. (2005). The phenomenon of distribution in

software development projects: A taxonomy proposal.

In Proc. European Mediterranean Conf. Information

Systems. Citeseer.

Gumm, D. C. (2006). Distribution dimensions in software

development projects: A taxonomy. IEEE software,

23(5):45–51.

Holmstrom, H., Conch

uir, E.

O., Agerfalk, J., and Fitzger-

ald, B. (2006). Global software development chal-

lenges: A case study on temporal, geographical and

socio-cultural distance. In Global Software Engineer-

ing, 2006. ICGSE’06. International Conference on,

pages 3–11. IEEE.

Hosni, M., Idri, A., and Abran, A. (2017). Investigating

heterogeneous ensembles with ﬁlter feature selection

for software effort estimation. In Proceedings of the

27th International Workshop on Software Measure-

ment and 12th International Conference on Software

Process and Product Measurement, pages 207–220.

ACM.

Hossain, E., Babar, M. A., and Paik, H.-y. (2009). Us-

ing scrum in global software development: a system-

atic literature review. In Global Software Engineer-

ing (ICGSE), 2009 Fourth IEEE International Con-

ference, pages 175–184. IEEE Computer Society.

Idri, A., Abnane, I., and Abran, A. (2016a). Missing data

techniques in analogy-based software development ef-

fort estimation. Journal of Systems and Software,

117:595–611.

Idri, A. and Abran, A. (2001). A fuzzy logic based set of

measures for software project similarity: validation

and possible improvements. In Proceedings Seventh

International Software Metrics Symposium, pages 85–

96. IEEE.

Idri, A., Abran, A., and Khoshgoftaar, T. M. (2002a). Es-

timating software project effort by analogy based on

linguistic values. In Proceedings Eighth IEEE Sym-

posium on Software Metrics, pages 21–30. IEEE.

Idri, A. and Amazal, F. A. (2012). Software cost estima-

tion by fuzzy analogy for isbsg repository. Uncer-

tainty Modeling in Knowledge Engineering and De-

cision Making, pages 863–868.

Idri, A., Amazal, F. A., and Abran, A. (2016b). Accuracy

comparison of analogy-based software development

effort estimation techniques. International Journal of

Intelligent Systems, 31(2):128–152.

Idri, A., Hosni, M., and Abran, A. (2016c). Improved esti-

mation of software development effort using classical

and fuzzy analogy ensembles. Applied Soft Comput-

ing, 49:990–1019.

Idri, A., Hosni, M., and Abran, A. (2016d). Systematic lit-

erature review of ensemble effort estimation. Journal

of Systems and Software, 118:151–175.

Idri, A., Khoshgoftaar, T. M., and Abran, A. (2002b). In-

vestigating soft computing in case-based reasoning for

software cost estimation. International Journal of En-

gineering Intelligent Systems for Electrical Engineer-

ing and Communications, 10(3):147–157.

Jaakkola, H. (2009). Towards a globalized software indus-

try. Acta Polytechnica Hungarica, 6(5):69–84.

Kwasnik, B. H. (1999). The role of classiﬁcation in knowl-

edge representation and discovery.

Lamersdorf, A., Munch, J., Fernandez-del Viso Torre, A.,

Sanchez, C. R., and Rombach, D. (2010). Estimating

the effort overhead in global software development.

In Global Software Engineering (ICGSE), 2010 5th

IEEE International Conference on, pages 267–276.

IEEE.

Laurent, P., Mader, P., Cleland-Huang, J., and Steele, A.

(2010). A taxonomy and visual notation for mod-

eling globally distributed requirements engineering

projects. In 2010 5th IEEE International Conference

on Global Software Engineering, pages 35–44. IEEE.

Peixoto, Lima, C. E., Audy, J. L. N., and Prikladnicki,

R. (2010). Effort estimation in global software de-

velopment projects: Preliminary results from a sur-

vey. In Global Software Engineering (ICGSE), 2010

5th IEEE International Conference, pages 123–127.

IEEE.

Ramasubbu, N., Cataldo, M., Balan, R. K., and Herbsleb,

J. D. (2011). Conﬁguring global software teams: a

multi-company analysis of project productivity, qual-

ity, and proﬁts. In Software Engineering (ICSE), 2011

33rd International Conference on, pages 261–270.

IEEE.

ICSOFT 2019 - 14th International Conference on Software Technologies

224

Smite, D., Wohlin, C., Galvin¸a, Z., and Prikladnicki, R.

(2014). An empirically based terminology and taxon-

omy for global software engineering. Empirical Soft-

ware Engineering, 19(1):105–153.

Smite, D., Wohlin, C., Gorschek, T., and Feldt, R. (2010).

Empirical evidence in global software engineering: a

systematic review. Empirical software engineering,

15(1):91–118.

Usman, M., Britto, R., B

orstler, J., and Mendes, E. (2017).

Taxonomies in software engineering: A systematic

mapping study and a revised taxonomy develop-

ment method. Information and Software Technology,

85:43–59.

A Software Cost Estimation Taxonomy for Global Software Development Projects

225