Modeling Decision Structures Part 2: Question Charts (Q Charts™) and Hybrid Diagrams
Extracted from Decision Analysis — A Primer: How to Use DecisionSpeak™ and Question Charts (Q-Charts™), by Ronald G. Ross, 2013, available (free) on http://www.brsolutions.com/IPSpeakPrimers
In DecisionSpeak, a Question Chart (Q-Chart™ for short) is a graphic representation of two or more operational business decisions indicating how they are related (dependent). A Q-Chart enables practitioners to capture, visualize, and analyze the structure (decision structure) of related operational business decisions in declarative form.
Q-Charts serve a very different purpose than business process models. Q-Charts comprise
- Q-COEs, which capture the basic elements of operational business decisions. Each Q-COE™ indicates what question ("Q") is being asked, and usually one or more of the following: considerations ("C"), outcomes ("O"), and exceptions ("E").
- decision dependencies, which show how Q-COEs relate logically. As discussed in last month's column, decision dependencies fall into three categories.
Operational business decisions often prove much more complex than they first seem(!). Developing a Q-Chart is an important step before jumping with both feet into decision tables. Q-Charts also focus practitioners on scope, always a critical issue in business analysis.Figure 1 presents a typical Q-Chart. Note the vertical arrangement of dependency connectors and Q-COEs. Since many notations for modeling process show flow or sequence horizontally, DecisionSpeak always orients Q-Charts vertically.
Figure 1. Q-Chart for Deciding How Much to Charge for Shipping an Order
Observations about the Q-Chart illustrated in Figure 1:
- The highest Q-COE, representing the most dependent decision, represents the question What should be charged for shipping an order? Four considerations are used in determining the outcome shipping cost.
- A consideration dependency connects the top Q-COE with the Q-COE beneath it What method should be used to ship an order? The outcome of the lower Q-COE provides the consideration kind of service for the upper (more dependent) Q-COE.
- The Q-COE What method should be used to ship an order? is connected via another consideration dependency to the Q-COE beneath it What is the priority level of a customer? The outcome of the lower Q-COE provides the consideration priority level of customer for the upper (more dependent) Q-COE.
- The Q-COE What method should be used to ship an order? is also connected via a relevance dependency to the Q-COE (lower left) Can an order be shipped? If an order cannot be shipped, it is meaningless to ask How should an order be shipped?
- The Q-COE Can an order be shipped? has four independent subdecisions beneath it, each presumably based on its own considerations.
- Finally, an outcome dependency connects the top Q-COE with the Q-COE beneath it (middle right) What set charges are there for shipping an order? This lower Q-COE will dictate certain outcomes for the upper (dependent) Q-COE.
Hybrid Dependency Diagrams
A proper Q-Chart includes only hexagons (Q-COEs), each representing a distinct question. The Q-COEs implicitly indicate probable links to external decision tables (often not yet developed).
Other forms of dependency diagrams may be used to show logical dependencies between business rules — i.e., where the expression of one business rule references something computed or derived by another business rule. Two simple examples:
- Mathematical dependency: The expression of one business rule references age, which is computed by another (computation) business rule based on birthdate.
- Derivation dependency: The expression of one business rule references gold customer. Whether a customer is gold or not is derived (determined) by another (derivation) business rule based on a dollar threshold of orders shipped.
An operational business decision may be based on both
- computation or derivation rules for certain well-defined cases.
- subdecisions for all other cases in scope.
In such cases, a mixed or hybrid dependency diagram can prove very useful. Such a diagram can portray the fundamental structure of highly-complex decision logic in a single diagram.
A hybrid dependency diagram includes a mix of individual decision rules (represented by boxes) and subdecisions (shown as Q-COEs), all based on a single (complex) question (operational business decision). The goal is to represent the associated decision logic in a unified, highly-approachable form and to provide a visual aid to help readers absorb know-how that is intrinsically complex.
Figure 2. Hybrid Dependency Diagram
In Figure 2 the operational business decision (at the top) asks What is the best antigen series for the patient? Immediately beneath the question is a 'stub' for a derivation rule, An antigen series is the best patient series if … .
- On the left side of the diagram are four boxes indicating cases in which the correct or optimal answer can be determined directly (i.e., without reference to decision tables). The text in each box simply completes the rule 'stub'.
Collectively the four boxes can be viewed as either representing four individual business rules, or better, as a single business rule. Using RuleSpeak tabulation the unified business rule could be expressed as follows:
An antigen series must be considered the best patient series if any of the following is true:
- Only one antigen exists for that antigen series.
- Exactly one antigen series is completed for the patient.
- Exactly one antigen series is in-process for the patient, and no antigen series is completed for the patient.
- All the antigen series given to the patient do not have any valid dose, and the antigen series is the default for the antigen.
Specification of this business rule separately from the diagram, however, is not necessary — the left side of the diagram is already practicable as shown.
- On the right side of the diagram are three cases in which the correct or optimal answer cannot be determined directly — so the answer is based on subdecisions.
At the bottom of the diagram is a derivation rule given in RuleSpeak that provides the meaning for the special vocabulary Wins on Points.
These subdecisions (and the associated decision tables — not shown) determine the 'winner' based on the highest number of points scored.
# # #