Basic Description

"Data Mining" has become a buzz-word within the computer industry for extraction of knowledge or information from large databases. Often the databases are financial, but the technology extends to other sorts of information databases. Similar ideas have existed in the statistics community for about 15 years under the name of exploratory data analysis. The convergence of these ideas coupled with recent advances in storage technology and database structures offer an interesting, exciting new technology. The idea of data mining is to look for information which may reside in opportunistically collected databases. A classic example involves a large discount chain sales records database in which it was noted that there was a high correlation between sales of disposable diapers and beer. The conclusion was that many men on their way home would be asked by their wives to pick up disposable diapers. The man decided to buy a six pack of beer while he was stopped anyway. The discount chain moved the beer and snacks such as peanuts and pretzels next to the disposable diapers and increased sales on peanuts and pretzels by more that 27%.

This course will focus on bringing together aspects of the computing, graphical/visualization and statistical technologies into an integrated treatment. This course will deal with data mining, exploratory analysis and knowledge discovery, particularly as they relate to graphical tools. Topics will include foundations, classification and clustering, trend and deviation analysis, dependency derivation, graphical tools, integrated discovery systems, and next generation databases.

Lecture Notes

Lecture notes will be made available as the course progresses. Watch this page for further materials.

• Lecture 1 Overview of Statistical Data Mining (Postscript, 152KB)
• Lecture 2 Computational Complexity (Postscript, 154KB)
  • Tables from Lecture 2 (Postscript, 136KB)
  • Read Chapter 1 of Fayyad et al. (1996)
• Lecture 3 Density Estimation (Postscript, 315KB)
  • Graphics from Lecture 3 (HTML)
  • Read Chapters 2, 3, 4 from Fayyad et al. (1996). NOTE: John Elder, co-author of Chapter 4 will be speaking at the Statistics seminar on September 20, 1996
• Lecture 4 Clustering and Classification (Postscript, 564KB)
  • Graphics from Lecture 4 (GIFs)
• Lecture 5 Graphical Methods for Clustering (HTML)
• Lecture 6 Classification and Regression Trees
• Lecture 7 Genetic Algorithms
• Lecture 8 Cognostics and the Analysis of Large Data Sets
• Lecture 9 Clustering by Mode Forests, Time Series Structures
  • The Mode Forest Paper is available in postscript form. (Postscript, 1.5MB)
  • The Time Series Lecture Notes are available in postscript form. (Postscript, 12MB)
  • Read Chapters 7, 8, 9, 10 from Fayyad et al. (1996).
• Lecture 10 Time Domain Time Series Models
• Lecture 11 Forecasting and Multiple Time Series
  • Paper handouts will be provided for this lecture.
  • Read Chapters 19, 20, 21, 22, 23 from Fayyad et al. (1996).
• Lecture 12 Nonlinear Nonparametric Regression: Convolution Methods, LOESS, Ridges and Skeletons
  • Paper handouts will be provided for this lecture.
  • Browse Remaining Chapters from Fayyad et al. (1996).