Changeset 11809:cf2369b2427d in orange


Ignore:
Timestamp:
12/13/13 19:09:19 (4 months ago)
Author:
blaz <blaz.zupan@…>
Branch:
default
Message:

Updated documentation on Data Sampler widget.

Location:
docs/widgets/rst/data
Files:
3 added
2 edited

Legend:

Unmodified
Added
Removed
  • docs/widgets/rst/data/datasampler.rst

    r11778 r11809  
    55 
    66.. image:: ../../../../Orange/OrangeWidgets/Data/icons/DataSampler.svg 
     7   :alt: Data Sampler icon 
     8   :class: widget-category-data widget-icon 
    79 
    810Selects a subset of data instances from the input data set. 
     
    1214 
    1315Inputs: 
    14  
    15  
    16    - Examples (ExampleTable) 
    17       Attribute-valued data set. 
    18  
     16    - Data 
     17        Input data set to be sampled. 
    1918 
    2019Outputs: 
    21  
    22  
    23    - Sample (ExampleTable) 
    24       Attribute-valued data set as sampled from the input data. 
    25    - Remaining Examples (ExampleTable) 
    26       Data instances from input data set that are not included in the sampled data. 
    27  
     20    - Data Sample 
     21        A set of sampled data instances. 
     22    - Remaining Data 
     23        All other data instances from input data set that are not included 
     24        in the sample. 
    2825 
    2926Description 
    3027----------- 
    3128 
    32 Data Sampler supports provides support for several means of 
    33 sampling of the data from the input channel and outputs the sampled 
     29Data Sampler implements several means of 
     30sampling of the data from the input channel. It outputs the sampled 
    3431data set and complementary data set (with instances from the input set 
    3532that are not included in the sampled data set). Output is set when the 
    36 input data set is set to the widget or after :obj:`Sample Data` is 
     33input data set is provided and after :obj:`Sample Data` is 
    3734pressed. 
    3835 
    39 .. image:: images/DataSampler.png 
     36.. image:: images/DataSampler-stamped.png 
    4037   :alt: Data Sampler 
     38   :align: right 
    4139 
    42 Sampling may be stratified: if input data contains a class, 
    43 sampling will try to match its class distribution in the output data 
    44 sets. 
     40.. rst-class:: stamp-list 
    4541 
    46 Several types of sampling are supported. :obj:`Random 
    47 sampling` can draw a 
    48 fixed number of instances or create a data set with a size set as 
    49 a proportion of instances from the input data set. In repeated 
    50 sampling, an data instance may be included in a sampled data several 
    51 times (like in bootstrap). 
     42   1. Info on input and output data set. 
     43   #. If input data contains a class, sampling will try to match 
     44      its class distribution in the output data sets. 
     45   #. Set random seed to always obtain the same sample given a choice of 
     46      data set and sampling parameters. 
     47   #. :obj:`Random sampling` can draw a 
     48      fixed number of instances or create a data set with a size set as 
     49      a proportion of instances from the input data set. In repeated 
     50      sampling, an data instance may be included in a sampled data several 
     51      times (like in bootstrap). 
     52   #. :obj:`Cross validation`, :obj:`Leave-one-out` or sampling that creates 
     53      :obj:`Multiple subsets` of preset sample sizes relative to the input 
     54      data set (like random sampling) all create several data samples. 
     55      Cross validation would split the data to equally-sized subsets 
     56      (:obj:`Number of folds`), and consider one of these as a sample. 
     57      Leave-one-out randomly chooses one data instance; all other instances 
     58      go to :obj:`Remaining Data` channel. Multiple subsets can create subset 
     59      of different sizes. 
     60   #. For sampling methods that create different data subsets, this 
     61      determines which subset is pushed to the :obj:`Data Sample` channel. 
     62   #. Press :obj:`Sample Data` to push the sample to the output 
     63      channel of the widget. 
    5264 
    53 :obj:`Cross validation`, 
    54 :obj:`Leave-one-out` or sampling that creates 
    55 :obj:`Multiple subsets` of preset sample sizes relative to the input data set 
    56 (like random sampling) all create several data samples. Which one is 
    57 send to the output is determined by the data set index in :obj:`Fold/Group` 
    58 (indices start with 1). 
     65.. container:: clearer 
    5966 
    60 Examples 
    61 -------- 
     67   .. image :: images/spacer.png 
    6268 
    63 Schema where we have sampled 10 data instances from Iris data set 
    64 and presented this selection in Scatterplot widget is shown 
    65 below. 
     69Example 
     70------- 
    6671 
    67 .. image:: images/DataSampler-Example-S.gif 
    68    :alt: Schema with Data Sampler 
     72In the following workflow Schema where we have sampled 10 data instances 
     73from Iris data set and send original data and the sample 
     74to Scatterplot widget. Sampled data instances are plotted with filled circles. 
     75 
     76.. image:: images/DataSampler-Example.png 
     77   :alt: A workflow with Data Sampler 
Note: See TracChangeset for help on using the changeset viewer.