|
Term
|
Explanation
|
|
A
|
|
|
|
access permissions
|
Controls that regulate who can read or alter computer files or directories. Project portfolio management (PPM) tools typically use
access permissions to ensure that users can view information and take actions that are specific to their defined roles.
|
|
aggregation equation
|
An equation specifying the rules used within a decision model to combine value judgments and measures of performance to
yield an overall measure of the value of decision options. The aggregation equation must be consistent with the independencies and dependencies that exist in decision-maker
preferences.
|
|
agile
|
As used in agile software development, refers to software development practices and methodologies that promote evolving, timely solutions based on
frequent reassessment of requirements and interim deliverables by collaborating, cross-function teams involving customers, designers, and developers. Similarly, as used in
agile project management, refers to project management methodologies that emphasize collaboration, quick delivery time, and ability to respond to changing requirements.
Six sigma and PRINCE2 are examples of a methodologies that incorporate principles of agile project management.
|
|
algorithm
|
A procedure composed of a sequence of instructions or steps for solving a problem (typically a problem expressed mathematically).
|
|
analysis
|
A systematic approach to building understanding based on investigating the component parts of a whole and their relationships. The most powerful methods of analysis
involve applying theories and techniques developed within specific fields of expertise, such as those from the natural science, social science, engineering, decision science, logic, and
mathematics.
|
|
analytic hierarchy process (AHP)
|
A theory-based mathematical decision-making technique developed by Thomas Saaty in the early 1970s. Some project portfolio management
(PPM) tools use AHP for project prioritization. Because there are many variations to how it is applied, it is impossible to provide
general conclusions about AHP's effectiveness for prioritizing projects.
As originally defined by Saaty, AHP involves asking decision makers to express their preferences over pairs of alternatives with regard to specific objectives. For example, "With regard to improving financial performance, do you prefer project A or project B, and by how much?" Oftentimes, a
nine-point scale is recommended for specifying strength of preference (the points on the scale are defined qualitatively, for example, equal preference, moderate preference, strong
preference, etc.). Based on the expressed preferences, a mathematical procedure is used to derive a model for ranking alternatives. Saaty provided an axiomatic theory to support this
approach to applying AHP. The theory has been criticized based on the observation that AHP can produce "rank reversals," a result wherein the
addition of a new alternative, say a new project, can change the ranking of existing projects, even though the new project does not influence the costs or benefits of the existing
projects. However, this undesirable result rarely occurs in practice.
Applications to project portfolio management generally use a variation of AHP in which the preference comparisons are expressed for objectives, not for the projects.
For example, "Compare the relative importance of the objectives 'improve time to market' and 'improve financial performance.'" The answers are used to derive a set of weights
interpreted to represent the relative importance of the objectives. Projects are then scored to indicate their contributions to each objective (e.g., no contribution = 0, slight
contribution = 0.1, ..., excellent contribution = 1). The scores are weighted and added to obtain an overall measure for ranking projects.
Like decision analysis, this version of AHP produces a model for evaluating projects. Since the model is derived by
encoding the fundamental preferences of the organization's decision makers, the underlying philosophy is that projects should be ranked based on how much they are preferred, not
according to goals such as "balance" or "strategic alignment." The decision model is simpler
than that typically produced by decision analysis in that it does not require the intermediate step of estimating the consequences of projects. Also, AHP normally does not bother with
assigning probabilities for uncertainties. Like traditional net present value analysis, all evaluations are conducted assuming a most likely scenario.
Unfortunately, this standard approach to applying AHP does not accurately prioritize projects (it also violates the axioms of Saaty's theory). It is not possible to
obtain meaningful preference comparisons between objectives unless the amounts of improvement are specified (e.g., the amount by which I prefer the objective "improve financial
performance" depends on how much financial performance is improved). Also, it is not correct to weight and add performance scores unless the value of achieving a given level of
performance on one measure does not depend on the level of performance achieved on any other measure, a condition known as preferential independence.
An alternative approach to applying AHP overcomes the above problems by using the step-by-step process prescribed by multi-attribute
utility analysis to ensure that weights refer to specified improvements against objectives and that performance measures satisfy the requirements necessary to permit them to be
weighted and added. In effect, the resulting approach to developing the model for valuing projects is multi-attribute utility analysis, however, AHP's pairwise comparison technique is
used to determine the weights.
|
|
analytics
|
The methods of analysis used to make optimal decisions based on data and mathematical logic. The term usually refers to the
application of more sophisticated forms of analysis. Project portfolio optimization is a common application of analytics.
|
|
application program
|
Any self-contained software program that performs a specific function directly for the user. This is in contrast to system software such as computer operating systems
that provide services to support application programs.
|
|
application programming interface (API)
|
An interface provided by project portfolio management tool or other application program that defines how that program
can access the computer's operating system or another program to request data or services.
|
|
asset management
|
The deliberate, long-term management of an organization's physical assets with the goal of maximizing their contribution to the achievement of the organization's
objectives. Asset management typically involves making decisions about when to create and acquire assets, how to use them, their repair or replacement, and their ultimate disposal.
Project portfolio management applied to asset intensive organizations often focuses on asset management and may be referred to as such.
|
|
attribute
|
A measure used to evaluate and compare decision alternatives in a multi-attribute utility analysis (MUA). An attribute quantifies the
degree to which an alternative achieves some decision objective. Properly defined attributes are measurable, operational, and understandable.
Measurable means that the attribute can serve as a measure. Specifically, for each alternative you could assign to the attribute a number (e.g., a point estimate) or, if there is uncertainty, a range or probability distribution. Operational
means that the attribute can be used to describe levels of achievement regarding the associated objective. Understandable means that there is no ambiguity in interpreting the meaning of
the attribute or the numbers that are assigned to it.
The definition of attributes is a critical step in the construction of a decision model based on MUA, as the choices made strongly affect the accuracy, defensibility,
practicality, and usefulness of the model. Attributes may be natural, constructed scales, or proxy measures. A natural measure is one that directly measures the achievement of the
corresponding objective in units that have wide and common usage. For example, cost, expressed in dollars (or euros, rubles, yen, etc.) is a natural measure for a cost objective.
A constructed scale is a scale that defines different levels for the attribute in terms of descriptions or definitions for each level of the scale. For example, while it might be
difficult to come up with a natural measure for the objective "improve corporate brand image," it might be possible to construct a 1-to-5 scale consisting of 5 different verbal
statements ranging from negative to positive customer perceptions of brand image. A proxy measure is an indirect measure selected because there is a presumed relationship that
exists between it and the objective. For example, if the company concerned about its image participates in a market analysis survey comparing customer perceptions, its ranking relative
to its competitors might serve as a proxy for its brand image.
|
|
audit trail
|
In the context of project portfolio management (non-accounting sense), evidence in the form of references, data or documents that enable a user to trace the path of
past actions, data changes, or decisions.
|
|
B
|
|
|
|
balanced scorecard
|
A process developed in the early 1990s by Robert Kaplan and David Norton for translating an organization's mission and strategy statements into a comprehensive system
for measuring organizational performance. Balanced scorecards collect diverse information intended to "balance" the traditional, but narrow, financial view of performance. The balanced
scorecard is an excellent tool for helping managers to understand how the organization is performing and helps translate strategy into action. However, balanced scorecards are not very
useful for prioritizing and choosing projects, and they are often misused in this regard.
According to the balanced-scorecard approach, performance measures should be defined in four areas: (1) finance, (2) customer satisfaction, (3) internal processes, and
(4) innovation and learning for employees. The selected measures are specific to the organization and are chosen to reflect the drivers believed to most important to understanding
success.
As examples, measures of organizational financial performance might include return on investment (ROI), rate of revenue growth,
amount of debt, etc. Customer satisfaction measures might include number of customer complaints, results of customer surveys, average time to process phone calls, etc. Internal process
measures might include fraction of projects delivered on schedule, number of units requiring rework, process yield rates, etc. Learning measures might include number of employee hours
spent in training, numbers of employee suggestions submitted, etc.
Measures can be backward-looking, to monitor how the organization is doing, or forward-looking, to assess the future impacts of alternative courses of actions.
Assessments against the measures are arrayed on pages or displays referred to as "scorecards." Target levels of performance may be assigned to the measures. Balanced scorecards are
being used in a broad range of activities, from product planning to incentive compensation, and by federal, state, and local governments.
The major weakness of balanced scorecards is that the approach does not provide a basis for trading off performance on different measures. In other words, if an
organization improves performance on one measure without degrading performance on any other measure, that is a good thing. However, if making a change intended to boost performance in a
given area (e.g., customer satisfaction) threatens performance in some other area (e.g., finance), a traditional balanced scorecard cannot indicate whether that change should be
made.
In an attempt to address the above weakness, balanced scorecards have sometimes been expanded to include a means for aggregating individual performance measures into a
quantity meant to represent the overall performance of the organization. For example, most commercially available tools for project portfolio management allow users to define equations
for combining performance measures—A scorecard is defined for assessing projects, and the various measures on the scorecard are mathematically aggregated to provide an indicator
intended to represent the relative desirability of the project. Typically, the form of the equation is weight-and-add (sometimes, the performance measures are merely averaged, which,
presumably, implies a desire to weight the measures equally).
Choosing projects so as to maximize a weighted sum of scorecard measures is nearly always incorrect. Yet, some try to justify this approach by arguing that
organizations should strive for balance in performance across various areas and measures. However, maximizing a weighted average does not necessarily lead to balance (if balance means
including projects that address all measures). In any case, the goal of project selection is to choose projects that create the most value, not balance (whatever balance means).
A project that has a high weighted-average performance score may or may not be a high-value project. How well the weighted-average score relates to value depends on,
among other things, how the measures are defined, the number of measures within each area and the degree to which they overlap or "double count," the organization's current level of
performance, and the basic objectives of the organization. It is possible to define performance measures that can be aggregated into an overall measure of project value, but this
requires a different process for defining performance measures than that used in the balanced scorecard approach (see multi-attribute utility analysis
(MUA).
|
|
Bayes theorem
|
A mathematical formula, originally developed in the 18th century by the minister and mathematician Thomas Bayes, that shows how probabilities should be updated based
on new information. The theorem is potentially applicable whenever actions are contemplated that would provide information relevant to decision making, and some project portfolio
management tools make use of Bayes theorem.
In its most basic form, Bayes theorem expresses the probability of some hypothesis (or outcome) H given that some event E occurs in terms of the initial (prior)
probability estimate of H and the likelihood (likelihood function) of E. Specifically:
where
- P(H) is the prior probability of H,
- P(H|E) is the conditional probability H given that event E is observed; and,
- P(E|H) and P(E|~H) is the likelihood function—it provides the conditional probability of E if H is and is not true.
For example, in the context of prioritizing tests for oil exploration, an important question would be the effectiveness of alternative tests at resolving uncertainty
over the presence of oil. In this case, the prior probability would be the initial estimate of the likelihood of striking oil at the location, for example 40%. The likelihood function
describes the accuracy of the test. For example, historically, the test may have signaled oil correctly 60% of the time when oil is present and incorrectly signaled oil 20% of the time
when oil is not present. If the test is conducted and if it signals oil, then, according to Bayes theorem, the posterior probability of oil would be: 0.6 x 0.4/(0.6 x 0.4 + 0.2 x 0.6) =
0.67. If data showed the test to be a more accurate indicator, for example, if 80% of the time it predicted oil when there was oil and only 10% predicted oil when there was no oil, the
test would have a more significant on posterior probability. With these numbers, the probability of oil given that the test predicts oil would be 0.94.
Bayes' Theorem provides a way to quantitatively describe the scientific method. If alternative hypotheses or models are competing for our
belief, we can test them by considering the consequences of each. We can then conduct tests to observe whether or not those consequences actually occur. If a hypothesis predicts
something should occur, and the something is a good indicator for the hypothesis, then the result strengthens our belief in the truthfulness of the hypothesis.
|
|
benefit
|
An increase in the achievement of a decision objective, for example, as a result of conducting a project. For any given
decision objective, conducting a project may cause benefits to increase (desirable), decrease (undesirable), or remain the same. Also, the benefits derived from a project can depend on
the other projects that are conducted (project interdependencies). To compare benefits across decision objectives it is necessary to measure benefits using a common unit. This is a goal
of multi-attribute utility analysis.
|
|
beta software
|
A new software product that is nearly fully developed but not yet thoroughly debugged. Beta versions of commercial application are often made available to customers
for free or at attractive prices, recognizing that there will likely be numerous problems such as crashes, errors, inconsistencies, etc.
|
|
