Software Engineering Body of Knowledge (v3) (2014) (811503), страница 60
Текст из файла (страница 60)
In such cases, plansdemand a review of initial goals as well as estimates and the underlying uncertainties and inaccuracies. Creative solutions with the underlyingrationale of achieving a win-win position areapplied to resolve conflicts.To be of value, planning should involve consideration of the project constraints and commitments to stakeholders.
Figure 12.4 shows howgoals are initially defined. Estimates are donebased on the initial goals. The plan tries to matchthe goals and the estimates. This is an iterativeprocess, because an initial estimate typically doesnot meet the initial goals.3.2. Estimation Techniques[3*, c6]Estimations are used to analyze and forecast theresources or time necessary to implement requirements (see Effort, Schedule, and Cost Estimationin the Software Engineering Management KAand Maintenance Cost Estimation in the SoftwareMaintenance KA). Five families of estimationtechniques exist:• Expert judgment• Analogy• Estimation by parts• Parametric methods• Statistical methods.No single estimation technique is perfect, sousing multiple estimation technique is useful.Convergence among the estimates produced bydifferent techniques indicates that the estimatesare probably accurate.
Spread among the estimates indicates that certain factors might havebeen overlooked. Finding the factors that causedthe spread and then reestimating again to produce results that converge could lead to a betterestimate.3.3. Addressing Uncertainty[3*, c6]Because of the many unknown factors duringproject initiation and planning, estimates areinherently uncertain; that uncertainty should beaddressed in business decisions.
Techniques foraddressing uncertainty include• consider ranges of estimates• analyze sensitivity to changes of assumptions• delay final decisions.3.4. Prioritization[3*, c6]Prioritization involves ranking alternatives basedon common criteria to deliver the best possiblevalue. In software engineering projects, softwarerequirements are often prioritized in order todeliver the most value to the client within constraints of schedule, budget, resources, and technology, or to provide for building product increments, where the first increments provide thehighest value to the customer (see RequirementsClassification and Requirements Negotiation inthe Software Requirements KA and SoftwareLife Cycle Models in the Software EngineeringProcess KA).3.5. Decisions under Risk[1*, c24] [3*, c9]Decisions under risk techniques are used whenthe decision maker can assign probabilities to thedifferent possible outcomes (see Risk Management in the Software Engineering ManagementKA).
The specific techniques include• expected value decision making• expectation variance and decision making• Monte Carlo analysis• decision trees• expected value of perfect information.12-12 SWEBOK® Guide V3.0Figure 12.5. The for-profit decision-making process3.6. Decisions under Uncertainty[1*, c25] [3*, c9]Decisions under uncertainty techniques are usedwhen the decision maker cannot assign probabilities to the different possible outcomes becauseneeded information is not available (see RiskManagement in the Software Engineering Management KA).
Specific techniques include• Laplace Rule• Maximin Rule• Maximax Rule• Hurwicz Rule• Minimax Regret Rule.4. Economic Analysis Methods4.1. For-Profit Decision Analysis[1*, c10]Figure 12.5 describes a process for identifyingthe best alternative from a set of mutually exclusive alternatives. Decision criteria depend on thebusiness objectives and typically include ROI(see section 4.3, Return on Investment) or Returnon Capital Employed (ROCE) (see section 4.4,Return on Capital Employed).For-profit decision techniques don’t apply forgovernment and nonprofit organizations.
In thesecases, organizations have different goals—whichmeans that a different set of decision techniquesare needed, such as cost-benefit or cost-effectiveness analysis.Software Engineering Economics 12-134.2. Minimum Acceptable Rate of Return[1*, c10]The minimum acceptable rate of return (MARR)is the lowest internal rate of return the organization would consider to be a good investment.Generally speaking, it wouldn’t be smart to investin an activity with a return of 10% when there’sanother activity that’s known to return 20%.The MARR is a statement that an organizationis confident it can achieve at least that rate ofreturn. The MARR represents the organization’sopportunity cost for investments. By choosingto invest in some activity, the organization isexplicitly deciding to not invest that same moneysomewhere else. If the organization is alreadyconfident it can get some known rate of return,other alternatives should be chosen only if theirrate of return is at least that high.
A simple wayto account for that opportunity cost is to use theMARR as the interest rate in business decisions.An alternative’s present worth evaluated at theMARR shows how much more or less (in present-day cash terms) that alternative is worth thaninvesting at the MARR.4.3. Return on Investment[1*, c10]Return on investment (ROI) is a measure of theprofitability of a company or business unit. Itis defined as the ratio of money gained or lost(whether realized or unrealized) on an investmentrelative to the amount of money invested. Thepurpose of ROI varies and includes, for instance,providing a rationale for future investments andacquisition decisions.4.4. Return on Capital EmployedThe return on capital employed (ROCE) is a measure of the profitability of a company or businessunit.
It is defined as the ratio of a gross profitbefore taxes and interest (EBIT) to the total assetsminus current liabilities. It describes the return onthe used capital.4.5. Cost-Benefit Analysis[1*, c18]Cost-benefit analysis is one of the most widelyused methods for evaluating individual proposals.
Any proposal with a benefit-cost ratio of lessthan 1.0 can usually be rejected without furtheranalysis because it would cost more than the benefit. Proposals with a higher ratio need to consider the associated risk of an investment andcompare the benefits with the option of investingthe money at a guaranteed interest rate (see section 4.2, Minimum Acceptable Rate of Return).4.6. Cost-Effectiveness Analysis[1*, c18]Cost-effectiveness analysis is similar to costbenefit analysis. There are two versions of costeffectiveness analysis: the fixed-cost versionmaximizes the benefit given some upper boundon cost; the fixed-effectiveness version minimizesthe cost needed to achieve a fixed goal.4.7. Break-Even Analysis[1*, c19]Break-even analysis identifies the point wherethe costs of developing a product and the revenueto be generated are equal. Such an analysis canbe used to choose between different proposals atdifferent estimated costs and revenue.
Given estimated costs and revenue of two or more proposals, break-even analysis helps in choosing amongthem.4.8. Business Case[1*, c3]The business case is the consolidated informationsummarizing and explaining a business proposalfrom different perspectives for a decision maker(cost, benefit, risk, and so on).
It is often usedto assess the potential value of a product, whichcan be used as a basis in the investment decisionmaking process. As opposed to a mere profitloss calculation, the business case is a “case” ofplans and analyses that is owned by the product12-14 SWEBOK® Guide V3.0manager and used in support of achieving thebusiness objectives.4.9. Multiple Attribute Evaluation[1*, c26]The topics discussed so far are used to make decisions based on a single decision criterion: money.The alternative with the best present worth, thebest ROI, and so forth is the one selected.
Asidefrom technical feasibility, money is almostalways the most important decision criterion, butit’s not always the only one. Quite often there areother criteria, other “attributes,” that need to beconsidered, and those attributes can’t be cast interms of money.
Multiple attribute decision techniques allow other, nonfinancial criteria to be factored into the decision.There are two families of multiple attributedecision techniques that differ in how they usethe attributes in the decision. One family is the“compensatory,” or single-dimensioned, techniques. This family collapses all of the attributesonto a single figure of merit. The family is calledcompensatory because, for any given alternative,a lower score in one attribute can be compensatedby—or traded off against—a higher score in otherattributes.
The compensatory techniques include• nondimensional scaling• additive weighting• analytic hierarchy process.In contrast, the other family is the “noncompensatory,” or fully dimensioned, techniques.This family does not allow tradeoffs among theattributes. Each attribute is treated as a separateentity in the decision process. The noncompensatory techniques include• dominance• satisficing• lexicography.4.10. Optimization Analysis[1*, c20]The typical use of optimization analysis is tostudy a cost function over a range of values tofind the point where overall performance is best.Software’s classic space-time tradeoff is anexample of optimization; an algorithm that runsfaster will often use more memory. Optimizationbalances the value of the faster runtime againstthe cost of the additional memory.Real options analysis can be used to quantifythe value of project choices, including the valueof delaying a decision.
Such options are difficultto compute with precision. However, awarenessthat choices have a monetary value providesinsight in the timing of decisions such as increasing project staff or lengthening time to market toimprove quality.5. Practical Considerations5.1. The “Good Enough” Principle[1*, c21]Often software engineering projects and productsare not precise about the targets that should beachieved. Software requirements are stated, butthe marginal value of adding a bit more functionality cannot be measured. The result could be latedelivery or too-high cost.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
