The starting point for this book is intelligence density. Intelligence density is a measure of the useful decision support information available.
Dhar and Stein posit a 4 quadrant stretch plot to facilitate the understanding:
– Quadrants:
– Q1: accuracy, explainability, response speed
– Q2: scalability, compactness, flexibility, embeddability, ease of use
– Q3: tolerance for complexity, tolerance for noise in data, toleance for sparse data, learning curve
– Q4: development speed, independence from experts, computing ease
– Dimensions:
– Q1 & Q2: Model Related
– Q3 & Q4: Organisation Related
– Q1 & Q3: Quality Related
– Q2 & Q4: Constraint Related
So, the 7 methods are:
– Data driven decision support
– Genetic algorithms: evolving solutions
– Neural networks: simulating the brain to solve problems
– rule-based systems: putting expert reasoning in a box
– fuzzy logic: dealing with linguistic ambiguity
– case-based reasoning: solving problems by analogy
– machine learning: deriving rules from data
As they highlight, utilising the correct method requires:
– an understanding of the range of tools & techniques available to model business problems
– a business-oriented methodology for developing decision support systems
For me, whilst all a bit theoretical and, at this stage, not applicable to my work situations, is an important piece of background information on the thinking behind business intelligence and decision support systems.
Leave a Reply