Changeset 9905:3fd7a62a81ee in orange


Ignore:
Timestamp:
02/07/12 12:02:51 (2 years ago)
Author:
tomazc <tomaz.curk@…>
Branch:
default
Children:
9906:77274e331dbb, 10031:3113e6606c8f
Message:

Minor changes to Orange.feature.imputation.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • docs/reference/rst/Orange.feature.imputation.rst

    r9890 r9905  
    282282capable of handling unknown values. 
    283283 
    284 Learners with imputer as a component 
    285 ==================================== 
     284Imputer as a component 
     285====================== 
    286286 
    287287Learners that cannot handle missing values should provide a slot 
     
    292292:obj:`~Orange.classification.logreg.LogRegLearner` will pass them to 
    293293:obj:`~Orange.classification.logreg.LogRegLearner.imputer_constructor` to get 
    294 an imputer and used it to impute the missing values in the learning data. 
    295 Imputed data is then used by the actual learning algorithm. Also, when a 
     294an imputer and use it to impute the missing values in the learning data. 
     295Imputed data is then used by the actual learning algorithm. When a 
    296296classifier :obj:`~Orange.classification.logreg.LogRegClassifier` is 
    297 constructed, 
    298 the imputer is stored in its attribute 
    299 :obj:`~Orange.classification.logreg.LogRegClassifier.imputer`. At 
    300 classification, the same imputer is used for imputation of missing values 
     297constructed, the imputer is stored in its attribute 
     298:obj:`~Orange.classification.logreg.LogRegClassifier.imputer`. During 
     299classification the same imputer is used for imputation of missing values 
    301300in (testing) examples. 
    302301 
     
    305304it is recommended to use imputation according to the described procedure. 
    306305 
    307 The choice of which imputer to use depends on the problem domain. In this 
    308 example we want to impute the minimal value of each feature. 
     306The choice of the imputer depends on the problem domain. In this example the 
     307minimal value of each feature is imputed: 
    309308 
    310309.. literalinclude:: code/imputation-logreg.py 
     
    318317.. note:: 
    319318 
    320    Note that just one instance of 
     319   Just one instance of 
    321320   :obj:`~Orange.classification.logreg.LogRegLearner` is constructed and then 
    322321   used twice in each fold. Once it is given the original instances as they 
     
    329328   testing. 
    330329 
    331 Wrapper for learning algorithms 
    332 =============================== 
     330Wrappers for learning 
     331===================== 
    333332 
    334333In a learning/classification process, imputation is needed on two occasions. 
    335 Before learning, the imputer needs to process the training examples. 
     334Before learning, the imputer needs to process the training instances. 
    336335Afterwards, the imputer is called for each instance to be classified. For 
    337336example, in cross validation, imputation should be done on training folds 
     
    343342simply skips the corresponding attributes in the formula, while 
    344343classification/regression trees have components for handling the missing 
    345 values in various ways. 
    346  
    347 If for any reason you want to use these algorithms to run on imputed data, 
    348 you can use this wrapper. 
     344values in various ways. A wrapper is provided for learning algorithms that 
     345require imputed data. 
    349346 
    350347.. class:: ImputeLearner 
Note: See TracChangeset for help on using the changeset viewer.