source: orange/orange/doc/widgets/Data/File.htm @ 9399:6bbe263e8bcf

Revision 9399:6bbe263e8bcf, 3.0 KB checked in by mitar, 2 years ago (diff)

Renaming widgets catalog.

Line 
1<html>
2<head>
3<title>File</title>
4<link rel=stylesheet href="../../../style.css" type="text/css" media=screen>
5</head>
6
7<body>
8
9<h1>File</h1>
10
11<p><img class="screenshot" src="../icons/File.png"/></p>
12<p>Reads the data from the input file.</p>
13
14<h2>Channels</h2>
15
16<h3>Inputs</h3>
17
18<DL class=attributes>
19<DT>(None)</DT>
20</dl>
21
22<h3>Outputs</h3>
23<DL class=attributes>
24<DT>Examples (ExampleTable)</DT>
25<DD>Attribute-valued data set read from the input file.</DD>
26</dl>
27
28<h2>Description</h2>
29
30<img class="leftscreenshot" src="File.png" align="left">
31
32<p>This is the widget you will probably use in every schema. It reads the
33input data file (data table with examples) and sends the data set to the
34output channel. It maintains a history of most recently used data files.
35For convenience, the history also includes a directory with the sample data
36sets that come with Orange.</p>
37
38<p>To browse for a data file press <span class="option">...</span>. <span class="option">Reload</span> button reloads the file.</p>
39
40<p>The widget can read data from simple tab-delimited or comma-separated files,
41as well as files in C4.5 format and Weka's .arrf files.</p>
42
43<p>The info box at the bottom of the widget displays information on the number of
44examples, attributes and meta attributes, and information on class
45variable (if defined in a file).</p>
46
47<p>The controls in between are for more advanced users.</p>
48
49<p>Tab-delimited file
50can have user defined symbols for undefined values. The symbols for "don't care"
51and "don't know" values can be specified in the corresponding edit lines. The default
52values for "don't know" and "don't care" depend upon format. Most users will use
53tab-delimited files: keep the field empty or put a question mark in there and that's
54it. Most algorithms do not differ between <em>don't know</em> and <em>don't care</em>
55values, so consider them both to mean <em>undefined</em>.</p>
56
57<p>The part which most will probably neither understand nor need is related to
58construction of new attributes. It becomes important when multiple data sets are
59loaded, usually with multiple File widgets. A typical example is loading separate
60training and testing data from two files. In this case Orange will usually treat
61the attributes with the same name but appearing in different files as the same
62attribute, so a classifier which uses the attribute "petal length" from the first
63will use the attribute of the same name from the second. In cases when attributes
64from different files just accidentally bear different names, one can instruct Orange
65to either always construct new attribute or construct them when they differ in
66their domains. Use this option with great care (if at all).</p>
67
68<h2>Examples</h2>
69
70<p>Any of your schemas should probably start with the File widget. In
71the schema below, the widget is used to read the data that is then
72sent to both <a href="DataTable.htm">Data Table</a> widget and to
73widget that displays <a
74href="../Visualize/AttributeStatistics">AttributeStatistics.htm</a>.</p>
75
76<img class="schema"
77src="File-Example.png" alt="Schema with File widget">
78
79</body>
80</html>
Note: See TracBrowser for help on using the repository browser.