Classification and Regression TreesRoutledge, 19 de out. de 2017 - 368 páginas The methodology used to construct tree structured rules is the focus of this monograph. Unlike many other statistical procedures, which moved from pencil and paper to calculators, this text's use of trees was unthinkable before computers. Both the practical and theoretical sides have been developed in the authors' study of tree methods. Classification and Regression Trees reflects these two sides, covering the use of trees as a data analysis method, and in a more mathematical framework, proving some of their fundamental properties. |
Conteúdo
MASS SPECTRA CLASSIFICATION | 203 |
72 Generalized Tree Construction | 205 |
REGRESSION TREES | 216 |
82 An Example | 217 |
83 Least Squares Regression | 221 |
84 Tree Structured Regression | 228 |
85 Pruning and Estimating | 232 |
86 A Simulated Example | 237 |
23 Construction of the Tree Classifier | 23 |
24 Initial Tree Growing Methodology | 27 |
25 Methodological Development | 36 |
26 Two Running Examples | 43 |
27 The Advantages of the Tree Structured Approach | 55 |
RIGHT SIZED TREES AND HONEST ESTIMATES | 59 |
32 Getting Ready to Prune | 63 |
33 Minimal CostComplexity Pruning | 66 |
An Estimation Problem | 72 |
35 Some Examples | 81 |
Appendix | 87 |
SPLITTING RULES | 93 |
41 Reducing Misclassification Cost | 94 |
42 The TwoClass Problem | 98 |
Unit Costs | 103 |
44 Priors and Variable Misclassification Costs | 112 |
45 Two Examples | 115 |
46 Class Probability Trees Via Gini | 121 |
Appendix | 126 |
STRENGTHENING AND INTERPRETING | 130 |
52 Variable Combinations | 131 |
53 Surrogate Splits and Their Uses | 140 |
54 Estimating WithinNode Cost | 150 |
55 Interpretation and Exploration | 155 |
56 Computational Efficiency | 163 |
57 Comparison of Accuracy with Other Methods | 168 |
Appendix | 171 |
MEDICAL DIAGNOSIS AND PROGNOSIS | 174 |
61 Prognosis After Heart Attack | 175 |
62 Diagnosing Heart Attacks | 182 |
63 Immunosuppression and the Diagnosis of Cancer | 189 |
64 Gait Analysis and the Detection of Outliers | 194 |
65 Related Work on ComputerAided Diagnosis | 201 |
87 Two CrossValidation Issues | 241 |
88 Standard Structure Trees | 247 |
89 Using Surrogate Splits | 248 |
810 Interpretation | 251 |
811 Least Absolute Deviation Regression | 255 |
812 Overall Conclusions | 264 |
BAYES RULES AND PARTITIONS | 266 |
92 Bayes Rule for a Partition | 269 |
93 Risk Reduction Splitting Rule | 272 |
94 Categorical Splits | 274 |
OPTIMAL PRUNING | 279 |
102 Optimally Pruned Subtrees | 284 |
103 An Explicit Optimal Pruning Algorithm | 293 |
CONSTRUCTION OF TREES FROM A LEARNING SAMPLE | 297 |
111 Estimated Bayes Rule for a Partition | 298 |
112 Empirical Risk Reduction Splitting Rule | 300 |
113 Optimal Pruning | 302 |
114 Test Samples | 303 |
115 CrossValidation | 306 |
116 Final Tree Selection | 309 |
117 Bootstrap Estimate of Overall Risk | 311 |
118 EndCut Preference | 313 |
CONSISTENCY | 318 |
121 Empirical Distributions | 319 |
122 Regression | 321 |
123 Classification | 324 |
124 Proofs for Selection 121 | 327 |
125 Proofs for Selection 122 | 332 |
126 Proofs for Selection 123 | 337 |
| 342 | |
| 347 | |
| 354 | |
Outras edições - Ver todos
Classification and Regression Trees Leo Breiman,Jerome Friedman,R.A. Olshen,Charles J. Stone Visualização parcial - 2017 |
Termos e frases comuns
a₁ accuracy algorithm B₁ Bayes rule best split bromine C(ij C₁ CART categorical variables Chapter Class 1 Node class probability estimation computed consists contains corresponding cross-validation estimates data sets decrease defined DEFINITION denote digit recognition distribution equal example Figure function Gini splitting given heart attack large number learning sample LEMMA linear combination linear regression M₁ mass spectra maximizes mean squared error measurement vectors method misclassification rate missing values nearest neighbor node impurity nonterminal node optimally pruned subtree P₁ partition patients percent predicted predictor prior probabilities priors procedure proof pruning algorithm R(t₁ random variables regression trees result risk Section sequence splitting rule standard error subsampling subsets Suppose surrogate splits t₁ T8 cells TABLE terminal nodes test sample estimates Theorem tion To(a tree construction tree grown tree selected tree structured tree structured classification variable importance variance waveform x₁