TY - JOUR AU1 - Fulkerson, Bill AB - BOOK REVIEWS In summary, I would recommend this book to technically advanced [. .I, is not recommended in general” (p. 175) Readers looking for a readers who need a good, practice-oriented reference on validation and “cookbook” with quick answers to technique selection may be frustrated. bootstrap. In a section entitled “Top Five Algorithms,” the author discusses a table of the “top five algorithms for all data sets.” He concludes that the “top Russell V. LENTH five most frequently [cited] areDIPOL92, ALLOCIO, Discrim, Logdiscr The University of Iowa and Quadisc, but too much should not be made of these figures as they depend very much on the mix of problems used,” (p. 185) More authors should be this objective in evaluating and interpreting their work! Which procedures are best? It depends on the data set! Based on a edited by D. MICHIE, D. J. SPIEGELHALTER, and C. C. correspondence analysis, neural networks and statistical procedures are TAYLOR, Hertfordshire, U.K.: Ellis Horwood, 1994, xiv + judged to do equally well on the same type of data sets. The scores 289 pp., $67.95. for neural network and statistical procedures are “virtually identical, but these [scores] are quite different from the score of TI - Machine Learning, Neural and Statistical Classification JF - Technometrics DO - 10.1080/00401706.1995.10484383 DA - 1995-11-01 UR - https://www.deepdyve.com/lp/taylor-francis/machine-learning-neural-and-statistical-classification-1dNcsS4qCk SP - 459 EP - 459 VL - 37 IS - 4 DP - DeepDyve ER -