BAYES

BAYES

Reverend Thomas Bayes  (1702-1761)  England mathematician

      first used probability inductively and established a mathematical basis for probability inference

      a Presbyterian minister in Tunbridge Wells from 1731

      Three published works on probability in 1731, 1736, 1763

      defended the views and philosophy of Sir Isaac Newton

      A fundamental proposition in probability was named after him, called Bayes Rule

      first used probability inductively and established a mathematical basis for probability inference

      a Presbyterian minister in Tunbridge Wells from 1731

      Three published works on probability in 1731, 1736, 1763

      defended the views and philosophy of Sir Isaac Newton

      A fundamental proposition in probability was named after him, called Bayes Rule

Bayesian Classification for Relational Databases

Bayesian Algorithem

      Provides a probability based solution for classification

      A direct method to find the best hypothesis from given training data

Implementation of Bayesian Algorithm

Features

      1. Test data – 20%

      2. New value in test data

n  Add into and update the training set

      3. Combine attribute values in groups

Advantage

Competitive to other classification

techniques

      Easy achievement

      Fast speed

      High stability

      Widely used

Disadvantage

      attribute independence assumption

   prediction accuracy of the inferred classifiers will not improve much when the training set size increases.

      Assumption accuracy

      lazy semi-naive Bayesian learning technique

      Adjusted probability naïve Bayesian induction 

      Bagging, Boosting, and Variants Naive Bayesian Learning

      Voting Classification Algorithms

Comment

      “The future of software may lie in the obscure theories of an 18th century cleric named Thomas Bayes.

 -   "Los Angeles Times", Business Section, October 28, 1995

      “Bayesian thinking is at the heart of many artificial intelligence programs including the Office 97 assistants.”

-  "PC @uthority Magazine", August 1998

      "...Gates said ‘Microsoft's competitive advantage was its expertise in Bayesian networks.' Is Gates onto something? Is this alien-sounding technology Microsoft's new secret weapon? Quite possibly."

-          "Los Angeles Times", Business Section, October 28, 1995

      There are more than 30 songs composed for Bayesian algorithm. The latest version of the songbook : http://www.biostat.umn.edu/~brad/songbook.pdf.