Data Mining. Concepts and Techniques, 3rd Edition
HAN 22-ind-673-708-9780123814791
Yüklə 7,95 Mb. Pdf görüntüsü
|
HAN 22-ind-673-708-9780123814791 2011/6/1 3:27 Page 696 #24 696 Index partitioning (Continued) recursive, 335 tuples, 334 Partitioning Around Medoids (PAM) algorithm, 455–457
partitioning methods, 448, 451–457, 491 centroid-based, 451–454 global optimality, 449 iterative relocation techniques, 448 k-means, 451–454 k-medoids, 454–457 k-modes, 454 object-based, 454–457 See also cluster analysis path-based similarity, 594 pattern analysis, in recommender systems, 282
pattern clustering, 308–310 pattern constraints, 297–300 pattern discovery, 601 pattern evaluation, 8 pattern evaluation measures, 267–271 all confidence , 268 comparison, 269–270 cosine, 268 Kulczynski, 268 max confidence , 268
null-invariant, 270–271 See also measures pattern space pruning, 295 pattern-based classification, 282, 318 pattern-based clustering, 282, 516 Pattern-Fusion, 302–307 characteristics, 304 core pattern, 304–305 initial pool, 306 iterative, 306 merging subpatterns, 306 shortcuts identification, 304
pattern-guided mining, 30 patterns actionable, 22 co-location, 319 colossal, 301–307, 320 combined significance, 312 constraint-based generation, 296–301 context modeling of, 314–315 core, 304–305 distance, 309 evaluation methods, 264–271 expected, 22 expressed, 309 frequent, 17 hidden meaning of, 314 interesting, 21–23, 33 metric space, 306–307 negative, 280, 291–294, 320 negatively correlated, 292, 293 rare, 280, 291–294, 320 redundancy between, 312 relative significance, 312 representative, 309 search space, 303 strongly negatively correlated, 292 structural, 282 type specification, 15–23 unexpected, 22
pattern-trees, 264 Pearson’s correlation coefficient, 222 percentiles, 48 perception-based classification (PBC), 348 illustrated, 349 as interactive visual approach, 607 pixel-oriented approach, 348–349 split screen, 349 tree comparison, 350 phylogenetic trees, 590 pivot (rotate) operation, 148 pixel-oriented visualization, 57 planning and analysis tools, 153 point queries, 216, 217, 220 pool-based approach, 433 positive correlation, 55, 56 positive tuples, 364 positively skewed data, 47 possibility theory, 428 posterior probability, 351 postpruning, 344–345, 346 power law distribution, 592 precision measure, 368–369 predicate sets frequent, 288–289 k, 289 predicates repeated, 288 variables, 295 prediction, 19 classification, 328 link, 593–594 loan payment, 608–609 with naive Bayesian classification, 353–355 numeric, 328, 385 HAN 22-ind-673-708-9780123814791 2011/6/1 3:27 Page 697 #25 Index 697 prediction cubes, 227–230, 235 example, 228–229 Probability-Based Ensemble, 229–230 predictive analysis, 18–19 predictive mining tasks, 15 predictive statistics, 24 predictors, 328 prepruning, 344, 346 prime relations contrasting classes, 175, 177 deriving, 174 target classes, 175, 177 principle components analysis (PCA), 100, 102–103 application of, 103 correlation-based clustering with, 511 illustrated, 103 in lower-dimensional space extraction, 578 procedure, 102–103 prior probability, 351 privacy-preserving data mining, 33, 621, 626 distributed, 622 k-anonymity method, 621–622 l-diversity method, 622 as mining trend, 624–625 randomization methods, 621 results effectiveness, downgrading, 622 probabilistic clusters, 502–503 probabilistic hierarchical clustering, 467–470 agglomerative clustering framework, 467, 469
algorithm, 470 drawbacks of using, 469–470 generative model, 467–469 interpretability, 469 understanding, 469
probabilistic model-based clustering, 497–508, 538 expectation-maximization algorithm, 505–508 fuzzy clusters and, 499–501 product reviews example, 498 user search intent example, 498 See also cluster analysis probability estimation techniques, 355 posterior, 351 prior, 351 probability and statistical theory, 601 Probability-Based Ensemble (PBE), 229–230 PROCLUS, 511 profiles, 614 proximity measures, 67 for binary attributes, 70–72 for nominal attributes, 68–70 for ordinal attributes, 74–75 proximity-based methods, 552, 560–567, 581 density-based, 564–567 distance-based, 561–562 effectiveness, 552 example, 552 grid-based, 562–564 types of, 552, 560 See also outlier detection pruning
cost complexity algorithm, 345 data space, 300–301 decision trees, 331, 344–347 in k-nearest neighbor classification, 425 network, 406–407 pattern space, 295, 297–300 pessimistic, 345 postpruning, 344–345, 346 prepruning, 344, 346 rule, 363 search space, 263, 301 sets, 345 shared dimensions, 205 sub-itemset, 263 pyramid algorithm, 101
quality control, 600 quantile plots, 51–52 quantile-quantile plots, 52 example, 53–54 illustrated, 53 See also graphic displays quantitative association rules, 281, 283, 288, 320 clustering-based mining, 290–291 data cube-based mining, 289–290 exceptional behavior disclosure, 291 mining, 289 quartiles, 48 first, 49 third, 49 queries, 10 intercuboid expansion, 223–225 intracuboid expansion, 221–223 language, 10 OLAP, 129, 130 point, 216, 217, 220 processing, 163–164, 218–227 range, 220 relational operations, 10
Yüklə 7,95 Mb. Dostları ilə paylaş:
|