|
Term
|
Explanation
|
|
funding pool
|
A source of funding available to an organization for conducting projects. The term is typically used when an organization has
multiple funding pools, for example, organizational accounts, customer accounts, or government grants. Funding pools typically establish constraints on the types of projects that may be
funded and the amounts that can be spent. For example, an airport may be able to use funds available from the Federal Aviation Administration (FAA) or city in which it is located to
help fund certain types of airport enhancement projects. Optimization methods may be used in project portfolio
management to construct value maximizing project portfolio while meeting the constraints established for each funding pool.
|
|
fuzzy logic
|
A method of mathematical reasoning that allows for vagueness and imprecision in the way that conditions are specified and in the conclusions that are drawn. Fuzzy
logic has been described as mimicking how humans make decisions, relying on natural language and rules as opposed to precise mathematical formulas. The approach has proven useful in a
very wide range of applications, and there are tools for project portfolio management based on fuzzy logic. However, despite its wide use, fuzzy logic
remains somewhat controversial, especially with regard to applications intended to support decision making.
Fuzzy logic was originally developed in the 1960's by engineering professor Lotfi Zadeh. Zadeh was working on the problem of how to represent within computers (which
use crisp, true-false, "one-zero" logic) vagueness in human language. For example, "Is a person of height 5' 10" tall?" Zadeh developed the concept of assigning a number between 0 and 1
to indicate the degree of truth in a statement. For example, with fuzzy logic, the statement "the person is tall" might be assigned the number 0.8, indicating a belief that the person
"mostly tall."
Zadeh and others have shown that the assignment of fuzzy "truth values" to statements allows for fuzziness in "if-then" rules and the derivation of what might be
called fuzzy conclusions. The traditional form of if-then logic is, "If A, then B"—precision in defining whether condition A is met leads to precision in the response B. With
fuzzy logic, one could stipulate that condition A is "mostly met," which would lead to a response of "mostly B." To illustrate, consider the example of an automated radar system for
enforcing the speed limit on roadways. With traditional logic, the system might be programmed to send you a ticket if its radar indicates you are going 26 miles per hour in a 25 mile
per hour zone. In contrast, a system based on fuzzy logic could conclude that you were mostly within the speed limit. Operating more like a real police officer, it might then mostly
decide not to issue you a ticket.
For most applications of fuzzy logic, the advantage is not that the method introduces vagueness into system response, but rather, that it allows for simpler and more
user-friendly designs for automated response systems, which often simplifies implementation and results in reduced product development costs. Thus, the largest application area for
fuzzy logic has been control systems, especially for devices such as consumer appliances where precision in response is not critical. For example, fuzzy logic has been used in the
design of control systems used in vacuum cleaners, microwave ovens, cameras, and washing machines. However, fuzzy logic has also been used in more critical decision support applications
such as medical diagnosis and pattern recognition.
Within project portfolio management, fuzzy logic has been applied in several different ways, including as a means for accounting for vagueness in the language used to
characterize candidate projects, to develop aggregate measures of project performance, and to emulate human reasoning.
Fuzzy logic controversy relates mainly to its perceived redundancy and/or inferiority relative to probability theory. Both approaches involve assigning numbers between
0 and 1 to propositions for the purpose of representing uncertainty. For example, each approach might assign a value of 0.75 to the statement "Sam is tall," however, the process and
interpretation would be very different. With probabilities, the analyst would first provide an unambiguous definition of "tall," for the purposes of the analysis, by specifying a
threshold expressed in feet and inches. The probability assignment would then represent uncertainty due to lack of complete knowledge about Sam's height. With fuzzy logic, imprecision
in the definition of tall is taken as given, and the assigned number represents that vagueness.
Critics of fuzzy logic claim that most, if not all applications of fuzzy logic could be accomplished as well, or more effectively, using probabilities. They point out,
for example, that fuzzy logic does not provide a basis for assigning the truth values that are input into fuzzy logic—the numbers must be assigned based on intuition. In this
respect, the approach can be argued to be inferior to subjective probabilities. Although subjective, probability numbers must be assigned with reference to an objective reality. A
probability of 0.75 can only be assigned if the outcome is believed to be exactly as likely as the chance of randomly drawing a red ball from an urn containing three white balls and one
red ball.
Despite criticisms, the success fuzzy logic has found in the world of practical applications demonstrates its usefulness as an analytic technique. From the standpoint
of project portfolio management, the main benefit of fuzzy logic is that it provides a means for capturing vagueness and imprecision.
|
|
G
|
|
|
|
GAAP
|
Stands for generally accepted accounting principles—a collection of rules, procedures and conventions, established by various government and private
organizations, to ensure that a company's financial statements accurately convey its financial status. Some vendors of project portfolio management
tools advertise that they incorporate metrics (or have other features) that facilitate compliance with GAAP.
|
|
game theory
|
A branch of applied mathematics that analyzes the behavior of individuals who are pursuing self interest against other individuals who are doing the same. Game theory
addresses situations in which an individual or organization chooses actions in situations where the consequences of those actions depend on the choices made by others. Applications of
game theory typically attempt to find an optimal strategy for one player to use when an opponent is also assumed to be playing optimally. Such strategies are termed equilibrium
strategies in that they are stable and unlikely to change. Game theory is used to support decision making in many different areas, including business and military strategy. Some project
portfolio management tool intended for application to investments in highly competitive environments utilize consequence models based on
game theory.
|
|
gaming
|
In project prioritization, the phenomenon wherein individuals deliberately bias estimates in order to improve the
evaluation of favored projects.
|
|
Gantt chart
|
A horizontal bar chart that displays the timing, duration, and interactions among multiple, time-phased activities, tasks, or projects. Gantt Charts, named after the originator Henry L Gantt, are a useful project management and
planning devise.
A simple Gantt chart
As shown in the example, a Gantt chart is a essentially a table with each row corresponding to an activity. Time (e.g., measured in days, weeks, or months) is denoted
by the columns. Each task is represented by a bar extending across the time columns, indicating the planned duration of the task. Milestones and critical path lines are used to add
further detail to the chart. Milestones are important checkpoints or deadlines and are indicated by small symbols in the time columns. Critical
path lines connect task bars to indicate dependencies, such as the requirement that a task be initiated after the completion or commencement of another task.
Ghant charting capability is routinely provided in tools for project management and in many project
portfolio management tools.
|
|
gap analysis
|
A systematic comparison of the current situation to the desired state. Gap analysis generally includes benchmarking and other
assessments for clarifying expectations. The end goal of gap analysis is the development of specific plan for closing the gap and moving organizational performance to the desired
state.
|
|
genetic algorithm
|
A mathematical search technique for solving optimization problems based on an algorithm
consisting of steps similar to those that occur in natural evolution. The technique has found applications in numerous fields, including biology, engineering, economics, and
manufacturing.
Basically, a genetic algorithm seeks an optimal solution by simulating an evolutionary process. An initial "population" of potential solutions is randomly generated.
Each candidate solution in this initial "generation" is evaluated to determine its "fitness." Those determined most fit "survive" and are combined and mutated to form a new population
(the next generation). The new population is then similarly used to conduct the next iteration of the algorithm. The process continues until either a maximum number of generations has
been produced or an acceptable fitness level has been obtained for a solution.
Flowchart of the basic genetic algorithm
Optimization based on generic algorithms is applicable to project portfolio management (ppm), and some ppm tools use the technique. In
this context, an initial generation of potential portfolios is randomly generated (parent portfolios). Portfolios that don't meet the constraints (e.g., cost constraints) are eliminated
(they die off). Pairs of individual portfolios are then combined (e.g., by choosing every other project from each) to produce second-generation portfolios (child portfolios). Child
portfolios that don't meet the constraints are eliminated. The remaining portfolios are then evaluated, and the highest ranked are selected to represent the next generation of
portfolios. This process is repeated for a set number of iterations or until the user-specified optimization parameters are satisfied.
A common problem with optimization algorithms is premature convergence—the optimizer homes in on a solution that is not really optimal. This occurs with genetic
algorithms because the population of potential solutions being used loses diversity. To address this problem, an approach is used that mimics the way nature maintains diversity. New
portfolios ("genetic mutations") are randomly generated and periodically introduced into the portfolio populations. The mutated portfolios only "survive" if they meet the constraints,
in which case they help keep the genetic algorithm from converging to a false optimum.
|
|
goal programming
|
An optimization method designed for problems with more than one objective that allows a solution to be found using linear programming.
The method, sometimes applied to resource allocation and project selection problems, involves expressing goals or targets for objectives and
assigning priorities or weights to achieving those targets. For example, the formulation might be to find the subset of R&D project opportunities that comes closest to achieving
specified goals for manpower utilization, market share, sales, and net present value maximization. Constraints on acceptable solutions may also be
defined, for example, requiring total costs to be no greater than the budget. Goal programming seeks solutions that meet constraints while minimizing the weighted sum of the deviations
from the specified targets. The solution effectively involves a repetitive process of attempting to achieve each goal, in order of priority, subject to the specified constraints.
Goal programming has several attractive features. One obviously, is the ability to address multiple objectives. Also, and importantly, solutions can be easily found
using the Simplex Method of linear programming. This means that relatively large numbers of decision variables, constraints, and goals may be established without creating
difficulties for finding a solution. For example, in the context of resource leveling, you could solve for multi-year solutions that
closely achieve many detailed goals with many specified constraints.
A feature that no doubt promotes the use of goal programming is the (apparent) non-demanding nature of the necessary inputs. Like other prioritization logics, a model
relating choices to performance is required. However, goal programming does not require detailed quantification of decision making preferences and willingness to make tradeoffs the way
that decision analysis methods do. Instead, all that is required is the specification of goal targets and weights.
The main disadvantage of goal programming, of course, is that reasonable goals and targets cannot be specified without reference to underlying decision-maker
preferences—choosing the "right" targets and weights is exceedingly difficult. If the targets and weights are not appropriate, the solution will not be the one in the best
interest of the organization. The reason for this is that goal programming does not allow tradeoffs between goals. For example, if sales growth is the first priority goal, and market
share is the second, the formulation implies that not even one dollar of sales growth can be sacrificed to obtain even a huge gain in market share. Goal programming represents a
"satisficing" approach to decision making, meaning that what is sought is a satisfactory solution rather than one that is truly optimal. Because the method does not capture willingness
to tradeoff achievement of the various objectives, it is incapable of finding solutions that lie on the efficient frontier.
|
|
Previous