Most of the recent research enterprises have focused on development of formal appraisal theoretical accounts to better appraisal truth. Formal appraisal theoretical accounts have been developed to mensurate lines of codification or size of package undertakings. But, most of the theoretical accounts have failed to better appraisal truth. This paper focuses on the reusability in package development attempt appraisal based on COCOMO II and ANN. Software reuse saves the package attempt and improves productiveness. Incorporation of reusability prosodies in COCOMO II may give better consequences. In COCOMO II it is really hard to happen the values of size parametric quantities. This paper has proposed a new theoretical account called COREANN ( COCOMO Reusability in Effort appraisal based on ANN theoretical account ) for better attempt appraisal truth and dependability. The proposed theoretical account has focused on two constituents of COCOMO II. First, alternatively of utilizing RUSE cost driver, three new reuse cost drivers are introduced. Second, In order to cut down the undertaking cost, three cost drivers such as PEXE, AEXE, LTEX are combined into individual cost driver Personnel Experience ( PLEX ) . Finally, this proposed theoretical account truth is more improved with the aid of ANN Enhanced RPROP algorithm and simulated tempering optimisation technique. To measure the public presentation of the proposed theoretical account a set of undertakings compared with the bing COCOMO II theoretical account by MRE, MMRE and PRED for rating of package cost appraisal. The concluding consequences show that the usage of proposed COREANN theoretical account attempt appraisal is dependable, accurate and predictable.

Cardinal words:

Reusability In Software Effort Estimation Model... TOPICS SPECIFICALLY FOR YOU

Effort Estimation, Software Reuse, COCOMO II, Artificial Neural Network, Simulated Annealing

1. Introduction

A study on package attempt appraisal revealed that most of the undertakings were failed due to attempt overproduction and transcending its original estimations. It besides stated that 60-80 per centum of package undertakings encounter attempt overproductions [ 1 ] . Effort overruns normally lead to be overproductions and missed undertaking deadline. This would do deficiency of productiveness or loss of concern. Software attempt appraisal is one of the most critical and complex undertaking in package technology, but it is an inevitable activity in the package development processes. Over the last three decennaries, a turning tendency has been observed in utilizing assortment of package attempt appraisal theoretical accounts in diversified package development processes. Along with this enormous growing, it is besides realized that the essentialness of all these theoretical accounts in gauging the package development costs and fixing the agendas more rapidly and easy in the awaited environments.

Although a great sum of research clip and money have been invested to better the truth of the assorted appraisal theoretical accounts. Due to the built-in uncertainness in package development undertakings such as complex and dynamic interaction factors, alteration of demands, intrinsic package complexness, force per unit area on standardisation and deficiency of package informations, it is unrealistic to anticipate really accurate attempt appraisal of package development processes [ 2 ] .

Software reuse has been given great importance in package development for decennaries. Software reuse has benefits such as decreased attempt, improved productiveness, decreased time-to-market and decreased cost. This research work addresses the significance of reusability in attempt appraisal and formulates new prosodies for reusability to find the dependable and accurate attempt estimations.

Predictable and dependable attempt appraisal is the disputing undertaking in package undertaking direction. In research there are Numberss of efforts to develop accurate cost appraisal theoretical accounts based on assorted techniques. However, the rating of truth and dependability of the theoretical account shows its advantages and failings. Choosing an appropriate theoretical account for a specific undertaking is an issue in project direction. The appropriate theoretical account which provides minimal comparative mistake has to be considered as the best tantrum for attempt appraisal. In last decennaries, assorted methods for cost and attempt appraisal have been proposed in the undermentioned three classs:

Adept Judgment ( EJ )

Algorithmic Models ( AM )

Machine Learning ( ML )

Adept Judgment ( EJ )

Expert judgement is used as an appraisal method [ 3 ] to measure the package attempt estimations. Inacurate consequences produced by the algorithmic and non-algorithmic cost theoretical accounts consequences in extended usage of EJ method to bring forth better estimations [ 4 ] .EJ estimations are strictly based on the sentiment provided by adept individuals of the organisation and it is the most often applied attempt appraisal technique [ 5 ] . The truth of estimation depends on the experience and expertness of the expert. Expert judgement based attempt appraisal is suited in unstable, altering environment and incorporate contextual information. Despite of its usefulness adept judgement have some drawbacks so the combination of theoretical account based adept judgement may bring forth better truth [ 2 ] .

1.2 Algorithmic Models ( AM )

Algorithmic theoretical accounts are popular for package attempt appraisal. When utilizing algorithmic attempt appraisal theoretical accounts, the cost driver is used to mensurate the attempt estimates. It maintains the relationship between attempt and one/more undertaking features. It needs standardization or to be adjusted to local fortunes. Examples of algorithmic theoretical accounts are the Constructive Cost Model ( COCOMO ) and the Software Lifecycle Management ( SLIM ) .

Slender Model:

The SLIM [ 6 ] an algorithmic estimating method was developed by Larry Putnam of Quantitative Software Management in 1970 ‘s. SLIM is practically used in big package undertakings and it is a tool for cost appraisal and adult male power scheduling. SLIM is fundamentally derived based on the construct of Norden-Rayleigh curve which evaluates manpower as a map of clip. SLIM depends on an Source Line of Code ( SLOC ) estimation for the undertaking ‘s general size, so modifies this through the usage of the Rayleigh curve theoretical account to bring forth its attempt estimations.

The SLIM package equation is determined as follows:

S= E * ( EFFORT ) 1/3 * td4/3

Where td is the package bringing clip,

Tocopherol is the environment factor that reflects the development capableness,

S is represented in LOC and

EFFORT is represent in person-year.

1.2.2 COCOMO II Model:

COCOMO II is an enhanced and updated theoretical account of COCOMO to carry through the demands of the following coevals package technology patterns [ 7 ] [ 8 ] . COCOMO II was published ab initio in the annals of package technology in 1995 with three bomber theoretical accounts ; an application-composition theoretical account, an early design theoretical account and a post-architecture theoretical account. COCOMO II has, as an input, a set of 17 Effort Multipliers ( EM ) or cost drivers which are used to set the nominal attempt ( PM ) to reflect the package merchandise being developed.

The Application Composition Model

The Application Composition theoretical account is proposed to gauge the attempt and agenda on undertakings that uses the latest package development tools which supports rapid application development. It purposes object points for sizing alternatively than the size of the codification. The capital size evaluate is fixed by numbering the figure of studies, screens and 3rd coevals constituents that will be utilized in application.

Effort ( PM ) = NOP/PROD

Where NOP = ( object points ) * ( 100- % reuse ) /100, PROD = NOP/PersonMonths.

The Early Design Model

This theoretical account is used acquire the rough estimations based on the preliminary probe and uncomplete undertaking analysis. It used to mensurate replacement package system architectures where UFP is used for sizing.

Effort = a*KLOC*EAF

Where a = 2.45, Effort Adjustment Factor ( EAF ) is computed as in original COCOMO theoretical account utilizing 7 prosodies such as RCPX, RUSE, PDIF, PERS, PREX, FCIL and SCED.

The Post Architecture Model

The Post-Architecture theoretical account is used when high degree design is completed and complete information about the undertaking is known. This theoretical account is used during the existent development of the undertaking or merchandise. This theoretical account consists of a set of 17 cost drivers and 5 scale factors.

Effort ( PM ) = a* ( SIZE ) b*EAF

Where a, B are package coefficients, a is set as 2.55 and B is computed as b=1.01+0.01* a?‘ ( Wisconsin ) , wi= amount of leaden graduated table factors.

1.3 Machine Learning ( ML ) :

Machine larning techniques have been used as an option to EJ and AM. Examples include fuzzed logic theoretical accounts, arrested development trees, Artificial Neural Networks ( ANN ) and instance based concluding [ 2 ] .

In earlier yearss of research, some research workers found that utilizing more than one technique can cut down hazard of dependability and better the truth and anticipations. Further, utilizing more than one method may avoid the loss of utile information that other methods can supply [ 9 ] . But, utilizing a combination of method seems to be a solution for supplying more trustable determinations in package attempt appraisal. Harmonizing to some study a combination of single methods has seldom been used to gauge package attempt. However it has been implemented successfully in other scientific Fieldss [ 9 ] .

1.3.1 Artificial Neural Network ( ANN ) Model:

The involvement on the application of ANN has grown in recent yearss. ANN has been applied to assorted job spheres such as technology, medical specialty, natural philosophies, etc. , ANN can be used as prognostic theoretical accounts because it is capable of patterning complex maps. ANN is a massively parallel computational theoretical account which simulates the belongingss of biological interrelated nerve cells. A nerve cell is the base of an ANN theoretical account which is described by a province, synapses, a combination map and a transportation map. The nerve cell computes a leaden amount of its inputs and generates an end product if the amount exceeds a certain threshold. This end product so becomes an input to other nerve cells in the web. The procedure continues until one or more end products are generated. Most of the package attempt appraisal techniques utilized back extension acquisition algorithm [ 10 ] [ 11 ] . The ANN is initialized with random weights and bit by bit learns the relationships implicit in a preparation informations set by seting its weights when presented to these informations.

Software reuse has been given great importance in package development for decennaries. Software reuse has benefits such as decreased attempt, improved productiveness, decreased time-to-market and decreased cost. This research work addresses the significance of reusability in attempt appraisal and formulates new prosodies for reusability to find the dependable and accurate attempt estimations. This paper proposed a new theoretical account called COREANN which is the enhanced reusability theoretical account of COCOMO II station architectural theoretical account. The truth of the proposed theoretical account has been improved with the aid of ANN Enhanced Resilient Backpropagation ( ERPROP ) algorithm and Simulated Annealing ( SA ) optimisation technique.

The remainder of the paper is organized as follows: Section 2 presents the reappraisal of the work done in the application of nervous web techniques to package attempt

appraisal. In this subdivision a brief description of back extension and RPROP are given. Section 3 describes the architecture of the proposed nervous web theoretical account. It besides discusses the acquisition algorithm used in developing the web. Section 4 describes the experimental apparatus i.e. , dataset readying and execution inside informations. Section 5 presents the experimental consequences obtained by using two different algorithms on the proposed architecture. Finally Section 6 summarizes our work and gives decisions and future research work.

2 Related Plants

2.1 Extensions of COCOMO II

The COCOMO II [ 7 ] [ 8 ] undertaking was started to run into the hereafter demands of the following coevals of package development procedure. The new COCOMO II theoretical account has incorporated characteristics that are realistic and accurate in COCOMO 81 and Ada COCOMO theoretical accounts. COCOMO II has proposed three submodels based on development phases of the undertaking. The Application Composition theoretical account is the first submodel used to gauge attempt and agenda on undertakings that use rapid application development tools. Early design theoretical account is used to acquire approximative estimation in the preliminary phases of the undertaking. Post architectural theoretical account is chiefly used to gauge attempt when the high degree design is completed. COCOMO II defined the reuse theoretical account which adjusts the codification reuse by modifying the size of the faculty or undertaking. This theoretical account considers reuse with map points and beginning lines of codification the same in either the early design theoretical account or the post-architecture theoretical account. A size estimation equivalent to the figure of lines of new beginning codification is computed and so adjusts the size estimation for new codification. This theoretical account has non clearly specified complete system to measure the “ existent ” tantamount SLOC. It is hard to graduate the theoretical account and hard to find the parametric quantities Design Modified ( DM ) , Code Modified ( CM ) , reuse package ( IM ) and Adapted SLOC.

Estimating development attempt utilizing reuse proposed by Balda and Gustafson [ 12 ] . This theoretical account adapted the simple COCOMO theoretical account by separating freshly developed codification that is specific to the undertaking, freshly developed codification that is made for reuse and codification that is modified for reuse. This theoretical account uses the four variables to stand for these types of codification.

COCOMO II Constructive Staged Schedule & A ; Effort Model ( COSSEMO ) [ 13 ] specifies the per centums of attempt and agenda to be applied to the different phases of undertaking: Origin, Elaboration and Construction. The predicted attempt and agenda from a COCOMO II correspond to the amount of attempt and agenda of origin, Elaboration and Construction phases. Therefore, the amount of the attempt or agenda for three phases can really number more than 100 % of the COCOMO II attempt and agenda.

Constructive RAD Schedule Estimation Model ( CORADMO ) [ 14 ] theoretical account has five drivers. Each driver has both evaluation degrees, which are selected by a user based on the features of the package undertaking, its development organisation, and its surroundings. There are numeral agenda and attempt multiplier values per phase for each evaluation degree. The impact of re-use of 3GL production codification is handled straight in the COCOMO II theoretical account via the re-use sub-model and its consequence on size. This CORADMO driver reflects the impact of re-use of codification and/or the usage of really high degree linguistic communications, particularly during the Inception and Elaboration phases. Higher evaluation degrees reflect the possible agenda compaction impacts in Inception and Elaboration phases due to faster prototyping, option geographic expedition. Clearly this impact will be dependent on the degree of capableness and experience in making this, such as Rapid Prototyping experience. The values of the multipliers matching to the evaluation degrees are the same for both attempt and agenda ; this implies that the staff degree is held changeless.

Constructive Quality Model ( COQUALMO ) [ 15 ] is an extension of the bing COCOMO II theoretical account to stipulate the quality. This theoretical account is based on the package defect debut and removal theoretical account described by Barry Boehm. The defects conceptually flow into a keeping armored combat vehicle through assorted defect beginning pipes. These defect beginning pipes are modeled in COQUALMO as the “ Software Defect Introduction Model ” . The Defect Introduction and Defect Removal Sub-Models described above can be integrated to the bing COCOMO II cost, attempt and agenda appraisal theoretical account.

Constructive COTS integrating cost theoretical account ( COCOTS ) [ 16 ] where COTS in bend is short for commercial-off-the-shelf, and refers to those pre-built, commercially available package constituents that are going of all time more of import in the creative activity of new package systems. This theoretical account was developed as an extension of the COCOMO II cost theoretical account for reclaimable constituents based package development attempt appraisal. COCOTS attempts to foretell the lifecycle costs of utilizing COTS constituents by capturing the more important COTS hazards in its mold parametric quantities.

The primary attack modeled by COCOMO is the usage of system constituents that are developed from abrasion or new codification. But COCOMO II besides allows you to pattern the reusability in which system constituents are built out of preexistent beginning codification. Even most the undertakings are non constructing the reuse constituent from abrasion but reclaimable constituent ‘s beginning codification can be modified to accommodate your demands. COCOMO II presently does non pattern the instance in which undertaking has entree to a preexistent constituent ‘s beginning codification.

2.2 ANN based Effort Estimation

Literature reveals that many package technology research workers have proposed ANN based attack to gauge package development attempt [ 10, 11, 17, 18, 19 ] . The back extension trained multilayered provender frontward networks is by and large used in most of the research work to foretell the package attempt appraisal. The usage of ANN with a back extension acquisition algorithm for attempt appraisal has explored [ 10,20,21 ] and found the effectivity of the nervous web technique in attempt appraisal. Some preliminary probe in the usage of nervous web in gauging package cost and produced really accurate consequences [ 19 ] , but the major set back in their work was due to the handiness of dataset and the truth of the consequence depends on the size of the preparation set.

3. Problem Statement

The chief and of import undertaking in the package development procedure is to calculate accurate and dependable attempt estimation. This appraisal should be realistic and trusted. Inaccurate estimates lead to major jobs in quality, agenda and cost. The possible failings in the current appraisal theoretical accounts motivate the demand for more accurate and realistic theoretical account for successful undertaking executing. But presents, package attempt appraisal theoretical accounts are inefficient for gauging attempt. The chief ground for most of the package attempt appraisal theoretical accounts failures are inability usage of reusability. Because package reuse has benefits such as decreased attempt, improved productiveness, decreased time-to-market and decreased cost. This paper focused chiefly on reusability in package attempt appraisal. This work identifies some of the drawbacks of COCOMO II attempt appraisal theoretical account in footings of reusability. To get the better of the drawbacks and ease better estimations, some of the cost drivers have been modified to back up reusability. This proposed theoretical account better the truth and to cut down an mistakes in attempt appraisal with the aid of ANN technique and optimise the solution utilizing Simulated Annealing algorithm. The attempt appraisal of proposed theoretical account is more accurate and dependable.

4. Proposed Model – COREANN

The major end of this proposed theoretical account is gauging more truth and dependable package attempt with the aid of package reusability construct. Comparing with COREANN, Software reusability in COCOMO II is non provided an truth consequence. Alternatively of RUSE cost driver, three new reuse cost drivers is introduced such as Reuse Veryhigh Level Language ( RVLL ) , Required Integrator for Product Reuse ( RIPR ) , Reuse Application Generator ( RAPG ) is giving best consequence for reusability in package attempt appraisal. The attempt appraisal expression of COREANN is,

— — -1


— — — — — — 2 — — — — — — — — — 2

The COREANN theoretical account Scale Factors are same as the COCOMO II [ 7 ] [ 8 ] theoretical account scale factors such as PREC, FLEX, RESL, TEAM, PMAT.

— — — — — — -3 — — — — — — — — — 3

— — — — — — 4 — — — — — — — — — 4

— 5

COREANN Cost Drivers:

Product dependability and complexness – RELY, DATA, CPLX, DOCU

Required reuse – RVLL, RIPR, RAPG

Platform trouble – Time, STOR, PVOL

Personnel capableness – ACAP, PCAP, PCON

Forces experience – PLEX

Facilities – TOOL, SITE

Required Development Schedule – SCED

4.1 New Metrics Introduction

Three cost drivers such as PEXE, AEXE, LTEX are combined into individual cost driver Personnel Experience ( PLEX ) for cut downing the package undertaking cost.

Alternatively of RUSE metric in COCOMO II, three new reuse prosodies are introduced,

RVLL ( Reuse Very high Level Language )

RIPR ( Required Integrator for Product Reuse )

RAPG ( Reuse Application Generator )

4.2 New Metrics Definition and Validation Methodologies

The Goal/Question/ Metric ( GQM ) paradigm provides a templet and guidelines to specify metric ends and polish them into concrete and realistic inquiries, which is later lead to the definition of steps. Software technology procedure requires feedback and rating mechanism to specify and formalize prosodies. GQM is useable as a practical guideline to plan and recycle technically sound and utile steps. It provides templets for specifying end and generate inquiries to specify new prosodies in package technology procedure [ 22 ] [ 23 ] .The chief focal point is to build cost drivers for prognostic theoretical accounts that set up a dependable attempt appraisal. Goals are defined in an operational manner by polishing them into a set of quantifiable inquiries that are used to pull out appropriate information. The new cost drivers are defined under GQM methodological analysis.

These new cost drivers are decently validated with the aid of Theoretical ( Internal ) proof and Empirical ( External ) proof [ 24 ] [ 25 ] . The of import of theoretical proof is to mensurate and asses the metric connotations utilizing DISTANCE model [ 26 ] and the empirical proof by garnering the information about the prosodies utilizing study method. To formalize the EAF of proposed theoretical account, company dataset incorporating 20 undertaking has been used. By seting the value of cost drivers, this will give better consequence than past undertakings.

4.3 COREANN with ANN Model Implementation:

To implement ANN theoretical account, COREANN attempt appraisal Equation 1 should be transform from non additive theoretical account to linear theoretical account by using natural logarithm on both sides. ANN is implemented with Enhanced RPROP.

In ( PM ) = In ( A ) + 0.91 * In ( SIZE ) + SF1 * 0.01 * In ( SIZE ) + aˆ¦aˆ¦aˆ¦ . + SF5 * 0.01 * In ( SIZE ) + In ( EM1 ) + In ( EM2 ) + aˆ¦aˆ¦aˆ¦ + In ( EM17 ) — — — — — — — – 6

[ Linear Equation ]

OPest =WT0 + WT1 * IP1 + WT2 * IP2 + aˆ¦+ WT6 * IP6 + WT7 * IP7 +aˆ¦+ WT23 * IP23

— — — — — — — — — — — – 7

[ ANN Based Model For Effort Estimation ]


OPest = In ( PM )

IP1 = 0.91 * In ( SIZE )

IP2 = SF1 * In ( SIZE ) , aˆ¦aˆ¦aˆ¦. , IP6 = SF5 * In ( SIZE )

IP7 = In ( EM1 ) , aˆ¦aˆ¦aˆ¦ , IP23 = In ( EM17 )

WT0 = In ( A )

WT1 = 1, aˆ¦aˆ¦aˆ¦aˆ¦. , WT23 = 1

IP1 to IP23 = & gt ; Inputs

OPest = & gt ; Output

WT0 = & gt ; Bias

WT1 aˆ¦aˆ¦ WT23 = & gt ; Weights ( Initial Value is 1 )

Actual ascertained attempt is compared with this estimated attempt. The differences between these values are the mistake in the attempt. It should be minimized.

4.4 Enhanced ERPROP Algorithm

The basic rule of ERPROP is to extinguish the harmful influence of the size of the partial derived function on the weight measure. Initially, the Enhanced RPROP algorithm is declared the undermentioned parametric quantities:

The addition factor value is I·+ = 1.2

The lessening factor value is I·- = 0.5

The initial update-value is a?†0 = 0.1 ( a?†ij = a?†0 )

The values of a?†max & lt ; 50 and a?†min & gt ; 1e-6

4.5. SA Optimization

In the proposed theoretical account COREANN, Simulated Annealing Algorithm [ 27 ] is used to gauge the optimal solution of the package undertaking attempt. The given solution method is helped to acquire optimum values of attempt:

— — — — — — — — — 8

Where, EffortM = Measured Value of Effort, EffortC = Computed Value of Effort harmonizing to the theoretical account used.

Fake Annealing Algorithm Procedure:

1. Low-level formatting: parametric quantities of tempering agenda.

2. Choose an loop mechanism: a simple prescription to bring forth a passage from current province to another province by a little disturbance.

3. Measure the new province, compute the value of i?„E = ( value of current province – value of new province ) .

4. If the new province is better, do it current province, otherwise probabilistically accept or reject it with a determined chance map

5. if status is true continue Step 2 otherwise terminated.

5. Performance Measures

Company database incorporating 20 undertakings is used to prove the proposed COREANN theoretical account. The undermentioned rating standard is used to measure and compare the public presentation of the proposed theoretical account with bing COCOMO II Model.

A common standard for the rating of cost appraisal theoretical account is the magnitude of comparative mistake ( MRE ) , and average magnitude of comparative mistake ( MMRE ) . MRE is defined as

— — — — — — — — — 9

And Mean Magnitude of Relative Error ( MMRE ) for N undertakings is defined as [ 11 ]

— — — — -10 — — — — — — — — — 10

Following to cipher the PRED ( P ) value. If lower MRE & A ; MMRE and higher PRED ( 25 ) , the package appraisal theoretical account attempt is more accurate and predictable than other theoretical accounts.

— — — — — — 11 — — — — — — — — — 21

where K is the figure of undertakings where MRE is less than or equal to p ( usually p value is 25 % ) .

5.1 Consequences

Out of the 20 undertaking dataset, to calculate an attempt of the proposed theoretical account. The estimated attempt is comparing with bing COCOMO II and Actual attempt of the undertaking. This consequences are shown as Table – 1 and comparing graph besides provided as below:

Table – 1: Comparison of Effort Estimation With SA Optimization

Table – 2 shows that the consequence for MRE comparing of the proposed theoretical account with bing COCOMO II.

Table 2 – Comparison of Effort Estimation Results In MRE

In Table – 2, MMRE value of COREANN is 17.019 and COCOMO II is 30.592. PRED ( 25 ) value of COREANN is 80.00 and COCOMO II is 35.00. By the above consequence, observed value for MMRE of COREANN is less than MMRE of COCOMO II and PRED ( 25 ) of COREANN is greater than PRED ( 25 ) of COCOMO II.

6. Decision and Future work

In package technology, it is highly hard to choose appropriate theoretical account for appraisal attempt appraisal due to the handiness of figure of theoretical accounts. Software reuse has become a major factor in development. Hence, attempt appraisal for reuse must accurate for the successful undertaking executing. This paper chiefly concentrated on the calculation of accurate attempt with package reusability as the chief focal point. While comparing public presentation consequences of COREANN and COCOMO II, it clearly shows that the proposed COREANN works better than COCOMO II. That is, the COREANN theoretical account is estimated lower MRE & A ; MMRE and higher PRED ( 25 ) than the COCOMO II theoretical account. So the anticipation truth of COREANN is high based on the public presentation avaluation. In future work, the attempt estimated by adept judgement method has to be considered to optimise the concluding attempt appraisal. Initial value of the optimisation is the attempt estimated by adept judgement.

Share this Post!

Kylie Garcia

Hi there, would you like to get such a paper? How about receiving a customized one?

Check it out