Written by: STATISTICA 9/29/2009 10:21 AM
I am in the middle of reading The Handbook of Statistical Analysis and Data Mining Applications by Robert Nisbet, John Elder and Gary Miner.
Disclosure: Because I consider Gary Miner a friend, I decided to read the book. I am very glad that I did...
This book has clarified some basic high-level thoughts. For example page eight and nine talks about TWO VIEWS OF REALITY. There is a brief recap of Aristotle vs Plato philosophy.
Aristotle defined reality based on what could be seen and touched. Reality was in the details. Plato was a big picture person who focused on ideas.
Why did I need to read about 2,000 year old philosophy?
I got my answer fairly quickly.
How do you understand complex systems? Do you start by understanding all the pieces of a process? Or do you start by understanding the big picture?
Does google.com, yahoo.com and bing.com really need to understand why one website was ranked as number one for a particular keyword? Not really.... They just need to understand their data mining algorithms and the error rates. They can take a big picture (Plato like) approach.
Here are a couple of quotes that go along with this thinking:
"In the 1980's, it became obvious to statistical mathematicians that the rigorously Aristotelian approach of the past was too restrictive for analyzing highly nonlinear relationships in large data sets in complex systems of the real world."
"The crushing practical needs of business to extract knowledge from data that could be leveraged immediately... required new analytical techniques.....Development of new techniques followed [a new path]... machine learning might be viewed as a blend of Aristotelian and Platonic approach to truth"
/aw
1 comment(s) so far...
Re: Data Mining, The Nature of Reality In point of fact the over emphasis of either the aristotelian or platonic approach is a distortion of reality, but a historically necessary one to a certain extent. The whole is always made up of parts. However, to obtain an understanding of how the parts fit and operate together to comprise the functioning whole, it (the whole) must be analyzed in detail.Reality is an interconnected whole. To understand reality we have to understand how the parts interconnect and interact to comprise a dynamic whole. Science is the analysis and synthesis of these interrelations and the discovery and study of the laws (dynamic patterns) which determine them.Thanks to the marvelous growth in techonology, especially the astronomical growth in storage and computing power, we can now begin to apprehend wholes without totally disrupting the interconnections and interrelationships. In a word, we are acquiring the ability (albeit yet at a very primitive stage) to analyze reality in "real time".
Re: Data Mining, The Nature of Reality
In point of fact the over emphasis of either the aristotelian or platonic approach is a distortion of reality, but a historically necessary one to a certain extent. The whole is always made up of parts. However, to obtain an understanding of how the parts fit and operate together to comprise the functioning whole, it (the whole) must be analyzed in detail.Reality is an interconnected whole. To understand reality we have to understand how the parts interconnect and interact to comprise a dynamic whole. Science is the analysis and synthesis of these interrelations and the discovery and study of the laws (dynamic patterns) which determine them.Thanks to the marvelous growth in techonology, especially the astronomical growth in storage and computing power, we can now begin to apprehend wholes without totally disrupting the interconnections and interrelationships. In a word, we are acquiring the ability (albeit yet at a very primitive stage) to analyze reality in "real time".