# source:orange/docs/widgets/rst/unsupervized/PCA.rst@10843:fb3d3ef6110e

Revision 10843:fb3d3ef6110e, 1.2 KB checked in by Ales Erjavec <ales.erjavec@…>, 23 months ago (diff)

Added rst documentation for the PCA widget.

Line
1.. _PCA:
2
3Principal Component Analysis
4============================
5
6.. image:: ../../../../Orange/OrangeWidgets/icons/Unknown.png
7    :alt: PCA
8
9Channels
10--------
11
12Inputs:
13    - Input Data (Table)
14
15Outputs:
16    - Transformed Data (Table)
17        PCA transformed input data.
18    - Eigen Vectors (Table)
19        Eigen vectors.
20
21
22Description
23-----------
24
25`Principal Component Analysis`_ (PCA) computes the PCA linear transformation
26of the input data.
27
28.. _`Principal Component Analysis`: http://en.wikipedia.org/wiki/Principal_component_analysis
29
30.. image:: images/PCA.png
31    :alt: PCA widget
32
33
34.. rst-class:: stamp-list
35
36    1. Specifies the maximum number of principal components to select
37       (a special value 'All' specifies all the components)
38    2. Specifies the number of components by the proportion of explained
39       variance
40    3. Output the transformed data and eigen vectors on any change to
41       the widget settings.
42    4. Send the transformed data and eigen vectors.
43    5. Graph of the explained variance by PCA componenets. The vertical
44       line can be dragged using the mouse.
45
46The number of components of the transformation can be selected using either
47the `Components Selection` input box or by dragging the vertical cutoff line
48in the graph (|5|).
Note: See TracBrowser for help on using the repository browser.