Data Mining: Introductory and Advanced Topics

Features

• Cover Type: Paperback with 315 pages
• Published by: Prentice Hall
• Edition: 1st Edition September 1, 2002
• Written in: English
• ISBN 10 Number: 0130888923
• ISBN 13 Number: 978-0130888921
• Book Dimensions: 9.1 x 7 x 0.9 inches
• Weighs: 1.2 pounds

Product Review



"It is the best book on data mining so far, and I would definitely adopt it for my course. The book is very comprehensive and covers all of the data mining topics and algorithms of which 1 am aware. The depth of coverage of each topic or method is exactly right and appropriate. Each algorithm is presented in pseudocode that is sufficient for any interested readers to convert into a working implementation in a computer language of their choice." - Michael H. Huhns, University of South Carolina





"Discussion on distributed, parallel, and incremental algorithms is outstanding." - Zoran Obradovic, Temple University



Back Cover Copy


Margaret Dunham offers the experienced data base professional or graduate level Computer Science student an introduction to the full spectrum of Data Mining concepts and algorithms. Using a database perspective throughout, Professor Dunham looks at algorithms, data structures, data types, and complexity of algorithms and space. This text emphasizes the use of data mining concepts in real-world applications with large database components. KEY FEATURES:

• Covers advanced topics such as Web Mining and Spatial/Temporal mining
• Includes succinct coverage of Data Warehousing, OLAP, Multidimensional Data, and Preprocessing
• Provides case studies
• Offers clearly written algorithms to better understand techniques
• Includes a reference on how to use Prototypes and DM products

Reader Reviews
Dunham gives a clear explanation of the main ideas in data mining. It's a concise book, directed towards the researcher or programmer. Space considerations meant that some topics are only briefly but succinctly covered, like fuzzy logic. More details are provided about neural networks, genetic algorithms and similarity measures. Bayesian classifications also get a good mention. Other classification measures involve distance-based methods to define clusters. For clustering, you should note that exactly what goes into a given cluster can be rather subjective. It could depend on your choice of metric. There is a fair amount of maths. Accessible to someone with a couple of years of university level maths, especially involving linear algebra. The section on Web mining is especially interesting. The Web is probably the largest database in the world. Certainly the most accessible. But with different characteristics from many other databases. Web data might be wrong, deliberately or otherwise. And some websites might be link farms, that try to pump up page rankings. Other databases simply don't have this concern about their contents. Dunham explains Google's PageRank and a competing idea from IBM. The algorithms are given in pseudocode. Which should not be a problem to an experienced programmer. Translating these into your choice of language is (or at least it should be) a lesser conceptual task than understanding the methods themselves. Or devising new methods. The book also aids the latter. Dunham's descriptions of the overall logic behind each algorithm is a good lead into what is needed in construction new ones.