source: orange/docs/widgets/rst/data/rank.rst @ 11810:60ae48329b9a

Revision 11810:60ae48329b9a, 2.2 KB checked in by blaz <blaz.zupan@…>, 4 months ago (diff)

Changed style in description of signals (for new widget documentation).

Line 
1.. _Rank:
2
3Rank
4====
5
6.. image:: ../../../../Orange/OrangeWidgets/Data/icons/Rank.svg
7   :class: widget-category-data widget-icon
8
9Ranking of attributes in classification or regression data sets.
10
11Signals
12-------
13
14Inputs:
15   - :obj:`Data`
16        Input data set.
17
18Outputs:
19   - :obj:`Reduced Data`
20        Data set which selected attributes.
21
22Description
23-----------
24
25Rank widget considers class-labeled data sets (classification or regression)
26and scores the attributes according to their correlation with the
27class.
28
29.. image:: images/Rank-stamped.png
30
31.. rst-class:: stamp-list
32
33   1. Attributes (rows) and their scores by different scoring methods
34      (columns).
35   #. Scoring techniques and their (optional) parameters.
36   #. For scoring techniques that require discrete attributes this is the number
37      of intervals to which continues attributes will be discretized to.
38   #. Number of decimals used in reporting the score.
39   #. Toggles the bar-based visualisation of the feature scores.
40   #. Adds a score table to the current report.
41
42Example: Attribute Ranking and Selection
43----------------------------------------
44
45Below we have used immediately after the :ref:`File`
46widget to reduce the set of data attribute and include only the most
47informative one:
48
49.. image:: images/Rank-Select-Schema.png
50
51Notice how the widget outputs a data set that includes only the best-scored
52attributes:
53
54.. image:: images/Rank-Select-Widgets.png
55
56Example: Feature Subset Selection for Machine Learning
57------------------------------------------------------
58
59Following is a bit more complicated example. In the workflow below we
60first split the data into training and test set. In the upper branch
61the training data passes through the Rank widget to select the most
62informative attributes, while in the lower branch there is no feature
63selection. Both feature selected and original data sets are passed to
64its own :ref:`Test Learners` widget, which develops a
65:ref:`Naive Bayes <Naive Bayes>` classifier and scores it on a test set.
66
67.. image:: images/Rank-and-Test.png
68
69For data sets with many features and naive Bayesian classifier feature
70selection, as shown above, would often yield a better predictive accuracy.
Note: See TracBrowser for help on using the repository browser.