Changeset 7498:945c2a301da2 in orange
 Timestamp:
 02/04/11 18:23:07 (3 years ago)
 Branch:
 default
 Convert:
 d71d78d3f8eed7640729db849df29b1ec3cc4241
 File:

 1 edited
Legend:
 Unmodified
 Added
 Removed

orange/Orange/classification/bayes.py
r7495 r7498 175 175 All attributes can also be set as constructor parameters. 176 176 177 :var adjustTreshold: If set and the class is binary, the classifier's 178 threshold will be set as to optimize the classification accuracy. 179 The threshold is tuned by observing the probabilities predicted on 180 learning data. Setting it to True can increase the 181 accuracy considerably 182 :var m: m for mestimate. If set, mestimation of probabilities 183 will be used using :class:`orange.ProbabilityEstimatorConstructor_m` 184 This attribute is ignored if you also set estimatorConstructor. 185 :var estimatorConstructor: Probability estimator constructor for 186 prior class probabilities. Defaults to 187 :class:`orange.ProbabilityEstimatorConstructor_relative` 188 Setting this attribute disables the above described attribute m. 189 :var conditionalEstimatorConstructor: Probability estimator constructor 190 for conditional probabilities for discrete features. If omitted, 191 the estimator for prior probabilities will be used. 192 :var conditionalEstimatorConstructorContinuous: Probability estimator 193 constructor for conditional probabilities for continuous features. 194 Defaults to 195 :class:`orange.ConditionalProbabilityEstimatorConstructor_loess` 177 .. attribute:: adjustTreshold 178 179 If set and the class is binary, the classifier's 180 threshold will be set as to optimize the classification accuracy. 181 The threshold is tuned by observing the probabilities predicted on 182 learning data. Setting it to True can increase the 183 accuracy considerably 184 185 .. attribute:: m 186 187 m for mestimate. If set, mestimation of probabilities 188 will be used using :class:`orange.ProbabilityEstimatorConstructor_m` 189 This attribute is ignored if you also set estimatorConstructor. 190 191 .. attribute:: estimatorConstructor 192 193 Probability estimator constructor for 194 prior class probabilities. Defaults to 195 :class:`orange.ProbabilityEstimatorConstructor_relative` 196 Setting this attribute disables the above described attribute m. 197 198 .. attribute:: conditionalEstimatorConstructor 199 200 Probability estimator constructor 201 for conditional probabilities for discrete features. If omitted, 202 the estimator for prior probabilities will be used. 203 204 .. attribute:: conditionalEstimatorConstructorContinuous 205 206 Probability estimator constructor for conditional probabilities for 207 continuous features. Defaults to 208 :class:`orange.ConditionalProbabilityEstimatorConstructor_loess` 196 209 """ 197 210 … … 253 266 :type baseClassifier: :class:`Orange.core.BayesLearner` 254 267 255 :var distribution: Stores probabilities of classes, i.e. p(C) for each 256 class C. 257 :var estimator: An object that returns a probability of class p(C) for a 258 given class C. 259 :var conditionalDistributions: A list of conditional probabilities. 260 :var conditionalEstimators: A list of estimators for conditional 261 probabilities 262 :var normalize: Tells whether the returned probabilities should be 263 normalized (default: True) 264 :var adjustThreshold: For binary classes, this tells the learner to 265 determine the optimal threshold probability according to 01 266 loss on the training set. For multiple class problems, it has 267 no effect. 268 .. attribute:: distribution 269 270 Stores probabilities of classes, i.e. p(C) for each class C. 271 272 .. attribute:: estimator 273 274 An object that returns a probability of class p(C) for a given class C. 275 276 .. attribute:: conditionalDistributions 277 278 A list of conditional probabilities. 279 280 .. attribute:: conditionalEstimators 281 282 A list of estimators for conditional probabilities 283 284 .. attribute:: adjustThreshold 285 286 For binary classes, this tells the learner to 287 determine the optimal threshold probability according to 01 288 loss on the training set. For multiple class problems, it has 289 no effect. 268 290 """ 269 291
Note: See TracChangeset
for help on using the changeset viewer.