| starting point | like a jigsaw puzzle jumbled in the box or crossword puzzle with the squares blank |
| well-defined goal | like the completed picture or crossword |
| allowable procedures | for moving from the starting point to the goal |
Someone who is skilled at solving a particular class
of puzzles has a good collection of procedures to go quickly and efficiently
from the starting point to the goal of most or all puzzles in the class.
These procedures range aross a spectrum from:
| algorithms | deterministic and guaranteed to succeed if given enough time | |
| to | heuristics & hunches | fallible but offer impressive savings in time and effort when they succeed |
The "puzzles" are more respectably referred to as "management science problems."
Mathematical programming problems start with
a goal function and a set of constraints and an optimal feasible
solution as the well-defined goal point; the puzzle is solved by answering
the question "What's best?" in terms of the assumptions of the problem
Time series forecasting problems start with
a series of numbers with dates, plus a set of assumptions (such as the
moving average assumptions), and the goal point is a formula which comes
closer to predicting the data than any other formula within the allowable
class of formulas
Simulation problems start
with a description of a real or potential system. The procedure is to develop
the description into a computer program representing the system. The program
inputs represent design & operation choices; and the program outputs
represent system performance. The goal point is reached by choosing program
inputs which will lead to outputs corresponding to desirable system performance
Decision analysis problems start with a collection
of allowable actions, a set of possible states of nature, and a utility
measure for each state-action pair. The goal point is to specify which
alternative action provides the best compromise between the hope for a
high utility and the fear of a low one