Data Mining: Practical Machine Learning Tools and Techniques, Second Edition
Yüklə 4,3 Mb. Pdf görüntüsü
|
5 1 6 I N D E X listOptions(), 482 literary mystery, 358 LMT, 408 load forecasting, 24–25 loan application, 22–23 local discretization, 297 locally weighted linear regression, 244, 251–253, 253–254, 323 locally weighted Naïve Bayes, 252–253
logic programs, 75 logistic model trees, 331 logistic regression, 121–125 LogitBoost, 328, 330, 331
logit transformation, 121 log-likelihood, 122–123, 276, 277 log-normal distribution, 268 log-odds distribution, 268
M5 ¢ program, 384 M5P, 408 M5Rules, 409 machine learning, 6 main(), 453 majority voting, 343 MakeDensityBasedClusterer, 419 MakeIndicator, 396, 398 makeTree(), 472, 480 Manhattan metric, 129 manufacturing processes, 28 margin, 324 margin curve, 324 market basket analysis, 27 market basket data, 55 marketing and sales, 26–28 Markov blanket, 278–279 Markov network, 283 massive datasets, 346–349 maximization, 265, 267 maximum margin hyperplane, 215–217
MDL metric, 277 MDL principle, 179–184 mean absolute error, 177–179 mean-squared error, 177, 178 measurement errors, 59 membership function, 121 memorization, 76 MergeTwoValues, 398 merging, 257 MetaCost, 319, 320
metadata, 51, 349, 350 metadata extraction, 353 metalearner, 332 metalearning algorithms in Weka, 414–418 metric tree, 136 minimum description length (MDL) principle, 179–184
miscellaneous classifiers in Weka, 405, 414 missing values, 58 classification rules, 201–202 decision tree, 63, 191–192 instance-based learning, 129 1R, 86
mixture model, 267–268 model tree, 246–247 statistical modeling, 92–94 mixed-attribute problem, 11 mixture model, 262–264, 266–268 MLnet, 38 ModelPerformanceChart, 431 model tree, 76, 77, 243–251 building the tree, 245 missing values, 246–247 nominal attributes, 246 pruning, 245–246 pseudocode, 247–250 regression tree induction, compared, 243 replicated subtree problem, 250 rules, 250–251 smoothing, 244, 251 splitting, 245, 247 what is it, 250 momentum, 233 monitoring, continuous, 28–29
P088407-INDEX.qxd 4/30/05 11:25 AM Page 516 I N D E X 5 1 7
multiclass alternating decision trees, 329, 330, 343
MultiClassClassifier, 418 multiclass learning problems, 334 MultilayerPerceptron, 411–413 multilayer perceptrons, 223–226, 231, 233 multinomial distribution, 95 multinomial Naïve Bayes, 95, 96 multiple linear regression, 326 multiresponse linear regression, 121, 124 multistage decision property, 102 multivariate decision trees, 199 MultiScheme, 417 myope, 13 N NaiveBayes, 403, 405 Naïve Bayes, 91, 278 clustering for classification, 337–338 co-training, 340 document classification, 94–96 limitations, 96–97 locally weighted, 252–253 multinomial, 95, 96 power, 96 scheme-specific attribute selection, 295–296 selective, 296 TAN (Tree Augmented Naïve Bayes), 279 what can go wrong, 91
nearest-neighbor learning, 78–79, 128–136, 235, 242 nested exceptions, 213 nested generalized exemplars, 239 network scoring, 277 network security, 357 neural networks, 39, 233, 235, 253 neural networks in Weka, 411–413
noise
data cleansing, 312 exemplars, 236–237 hand-labeled data, 338 robustness of learning algorithm, 306 noisy exemplars, 236–237 nominal attributes, 49, 50, 56–57, 119 Cobweb, 271 convert to numeric attributes, 304–305 decision tree, 62 mixture model, 267 model tree, 246 subset, 88 nominal quantities, 50
non-axis-parallel class boundaries, 242 Non-Bayesians, 141 nonlinear class boundaries, 217–219 NonSparseToSparse, 401 normal-distribution assumption, 92 normalization, 56
normalized expected cost, 175 nuclear family, 47 null hypothesis, 155 numeric attribute, 49, 50, 56–57 axis-parallel class boundaries, 242 classification rules, 202 Classit, 271 converting discrete attributes to, 304–305 decision tree, 62, 189–191 discretizing, 296–305. See also Discretizing numeric attributes instance-based learning, 128, 129 interval, 88 linear models, 119 linear ordering, 349 mixture model, 268 1R, 86
statistical modeling, 92 numeric-attribute problem, 11 numeric prediction, 43–45, 243–254 evaluation, 176–179 forward stagewise additive modeling, 325 linear regression, 119–120 locally weighted linear regression, 251–253 P088407-INDEX.qxd 4/30/05 11:25 AM Page 517 Yüklə 4,3 Mb. Dostları ilə paylaş:
|