source: orange/docs/widgets/rst/classify/c45.rst @ 11404:1a7b773d7c7b

Revision 11404:1a7b773d7c7b, 4.1 KB checked in by Ales Erjavec <ales.erjavec@…>, 13 months ago (diff)

Replaced the use of :code: role with :obj:

Line 
1.. _C4.5:
2
3C4.5 Learner
4============
5
6.. image:: ../icons/C4.5.png
7
8C4.5 learner by Ross Quinlan
9
10Signals
11-------
12
13Inputs:
14
15
16   - Examples (ExampleTable)
17      A table with training examples
18
19
20Outputs:
21
22   - Learner
23      The C4.5 learning algorithm with settings as specified in the dialog.
24
25   - Classifier
26      Trained C4.5 classifier
27
28   - C45 Tree (C45Classifier)
29      Induced tree in the original Quinlan's format
30
31   - Classification Tree (TreeClassifier)
32      Induced tree in Orange's native format
33
34
35:obj:`Classifier`, :obj:`C45 Tree` and :obj:`Classification Tree` are
36available only if examples are present on the input. Which of the latter two
37output signals is active is determined by setting
38:obj:`Convert to orange tree structure` (see the description below).
39
40Description
41-----------
42
43This widget provides a graphical interface to the well-known Quinlan's C4.5
44algorithm for construction of classification tree. Orange uses the original
45Quinlan's code which must be, due to copyright issues, built and linked in
46separately.
47
48Orange also implements its own classification tree induction algorithm which
49is comparable to Quinlan's, though the results may differ due to technical
50details. It is accessible in widget :ref:`Classification Tree`.
51
52As all widgets for classification, C4.5 widget provides learner and classifier
53on the output. Learner is a learning algorithm with settings as specified by
54the user. It can be fed into widgets for testing learners, namely
55:ref:`Test Learners`. Classifier is a classification tree build from the
56training examples on the input. If examples are not given, the widget outputs
57no classifier.
58
59.. image:: images/C4.5.png
60   :alt: C4.5 Widget
61
62Learner can be given a name under which it will appear in, say,
63:ref:`Test Learners`. The default name is "C4.5".
64
65The next block of options deals with splitting. C4.5 uses gain ratio by
66default; to override this, check :obj:`Use information gain instead of ratio`,
67which is equivalent to C4.5's command line option ``-g``. If you enable
68:obj:`Subsetting` (equivalent to ``-s``), C4.5 will merge values of
69multivalued discrete attributes instead of creating one branch for each node.
70:obj:`Probabilistic threshold for continuous attributes` (``-p``) makes
71C4.5 compute the lower and upper boundaries for values of continuous attributes
72for which the number of misclassified examples would be within one standard
73deviation from the base error.
74
75As for pruning, you can set the :obj:`Minimal number of examples in the leaves`
76(Quinlan's default is 2, but you may want to disable this for noiseless data),
77and the :obj:`Post prunning with confidence level`; the default confidence is
7825.
79
80Trees can be constructed iteratively, with ever larger number of examples. If
81enable, you can set the :obj:`Number of trials`, the
82:obj:`initial windows size` and :obj:`window increment`.
83
84The resulting classifier can be left in the original Quinlan's structure, as
85returned by his underlying code, or :obj:`Converted to orange the structure`
86that is used by Orange's tree induction algorithm. This setting decides which
87of the two signals that output the tree - :obj:`C45 Classifier` or
88:obj:`Tree Classifier` will be active. As Orange's structure is more general
89and can easily accommodate all the data that C4.5 tree needs for
90classification, we believe that the converted tree behave exactly the same as
91the original tree, so the results should not depend on this setting. You should
92therefore leave it enabled since only the converted trees can be shown in the
93tree displaying widgets.
94
95When you change one or more settings, you need to push :obj:`Apply`; this will
96put the new learner on the output and, if the training examples are given,
97construct a new classifier and output it as well.
98
99
100Examples
101--------
102
103There are two typical uses of this widget. First, you may want to induce the
104tree and see what it looks like, like in the schema on the right.
105
106.. image:: images/C4.5-SchemaClassifier2.png
107   :alt: C4.5 - Schema with a Classifier
108
109The second schema shows how to compare the results of C4.5 learner with another
110classifier, naive Bayesian Learner.
111
112.. image:: images/C4.5-SchemaLearner.png
113   :alt: C4.5 - Schema with a Learner
114
Note: See TracBrowser for help on using the repository browser.