| <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> |
| <html xmlns="http://www.w3.org/1999/xhtml"> |
| <head> |
| <meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/> |
| <meta http-equiv="X-UA-Compatible" content="IE=9"/> |
| <meta name="generator" content="Doxygen 1.8.4"/> |
| <title>MADlib: bayes.sql_in Source File</title> |
| <link href="tabs.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javascript" src="jquery.js"></script> |
| <script type="text/javascript" src="dynsections.js"></script> |
| <link href="navtree.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javascript" src="resize.js"></script> |
| <script type="text/javascript" src="navtree.js"></script> |
| <script type="text/javascript"> |
| $(document).ready(initResizable); |
| $(window).load(resizeHeight); |
| </script> |
| <link href="search/search.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javascript" src="search/search.js"></script> |
| <script type="text/javascript"> |
| $(document).ready(function() { searchBox.OnSelectItem(0); }); |
| </script> |
| <script type="text/x-mathjax-config"> |
| MathJax.Hub.Config({ |
| extensions: ["tex2jax.js", "TeX/AMSmath.js", "TeX/AMSsymbols.js"], |
| jax: ["input/TeX","output/HTML-CSS"], |
| }); |
| </script><script src="../mathjax/MathJax.js"></script> |
| <link href="doxygen.css" rel="stylesheet" type="text/css" /> |
| </head> |
| <body> |
| <div id="top"><!-- do not remove this div, it is closed by doxygen! --> |
| <div id="titlearea"> |
| <table cellspacing="0" cellpadding="0"> |
| <tbody> |
| <tr style="height: 56px;"> |
| <td style="padding-left: 0.5em;"> |
| <div id="projectname">MADlib |
|  <span id="projectnumber">1.1</span> <span style="font-size:10pt; font-style:italic"><a href="../latest/./bayes_8sql__in_source.html"> A newer version is available</a></span> |
| </div> |
| <div id="projectbrief">User Documentation</div> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <!-- end header part --> |
| <!-- Generated by Doxygen 1.8.4 --> |
| <script type="text/javascript"> |
| var searchBox = new SearchBox("searchBox", "search",false,'Search'); |
| </script> |
| <div id="navrow1" class="tabs"> |
| <ul class="tablist"> |
| <li><a href="index.html"><span>Main Page</span></a></li> |
| <li><a href="modules.html"><span>Modules</span></a></li> |
| <li class="current"><a href="files.html"><span>Files</span></a></li> |
| <li> |
| <div id="MSearchBox" class="MSearchBoxInactive"> |
| <span class="left"> |
| <img id="MSearchSelect" src="search/mag_sel.png" |
| onmouseover="return searchBox.OnSearchSelectShow()" |
| onmouseout="return searchBox.OnSearchSelectHide()" |
| alt=""/> |
| <input type="text" id="MSearchField" value="Search" accesskey="S" |
| onfocus="searchBox.OnSearchFieldFocus(true)" |
| onblur="searchBox.OnSearchFieldFocus(false)" |
| onkeyup="searchBox.OnSearchFieldChange(event)"/> |
| </span><span class="right"> |
| <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a> |
| </span> |
| </div> |
| </li> |
| </ul> |
| </div> |
| <div id="navrow2" class="tabs2"> |
| <ul class="tablist"> |
| <li><a href="files.html"><span>File List</span></a></li> |
| <li><a href="globals.html"><span>File Members</span></a></li> |
| </ul> |
| </div> |
| </div><!-- top --> |
| <div id="side-nav" class="ui-resizable side-nav-resizable"> |
| <div id="nav-tree"> |
| <div id="nav-tree-contents"> |
| <div id="nav-sync" class="sync"></div> |
| </div> |
| </div> |
| <div id="splitbar" style="-moz-user-select:none;" |
| class="ui-resizable-handle"> |
| </div> |
| </div> |
| <script type="text/javascript"> |
| $(document).ready(function(){initNavTree('bayes_8sql__in_source.html','');}); |
| </script> |
| <div id="doc-content"> |
| <!-- window showing the filter options --> |
| <div id="MSearchSelectWindow" |
| onmouseover="return searchBox.OnSearchSelectShow()" |
| onmouseout="return searchBox.OnSearchSelectHide()" |
| onkeydown="return searchBox.OnSearchSelectKey(event)"> |
| <a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(0)"><span class="SelectionMark"> </span>All</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(1)"><span class="SelectionMark"> </span>Files</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(2)"><span class="SelectionMark"> </span>Functions</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(3)"><span class="SelectionMark"> </span>Groups</a></div> |
| |
| <!-- iframe showing the search results (closed by default) --> |
| <div id="MSearchResultsWindow"> |
| <iframe src="javascript:void(0)" frameborder="0" |
| name="MSearchResults" id="MSearchResults"> |
| </iframe> |
| </div> |
| |
| <div class="header"> |
| <div class="headertitle"> |
| <div class="title">bayes.sql_in</div> </div> |
| </div><!--header--> |
| <div class="contents"> |
| <a href="bayes_8sql__in.html">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> <span class="comment">/* ----------------------------------------------------------------------- */</span><span class="comment">/**</span></div> |
| <div class="line"><a name="l00002"></a><span class="lineno"> 2</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment"> * @file bayes.sql_in</span></div> |
| <div class="line"><a name="l00004"></a><span class="lineno"> 4</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00005"></a><span class="lineno"> 5</span> <span class="comment"> * @brief SQL functions for naive Bayes</span></div> |
| <div class="line"><a name="l00006"></a><span class="lineno"> 6</span> <span class="comment"> * @date January 2011</span></div> |
| <div class="line"><a name="l00007"></a><span class="lineno"> 7</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00008"></a><span class="lineno"> 8</span> <span class="comment"> * @sa For a brief introduction to Naive Bayes Classification, see the module</span></div> |
| <div class="line"><a name="l00009"></a><span class="lineno"> 9</span> <span class="comment"> * description \ref grp_bayes.</span></div> |
| <div class="line"><a name="l00010"></a><span class="lineno"> 10</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00011"></a><span class="lineno"> 11</span> <span class="comment"> */</span><span class="comment">/* ----------------------------------------------------------------------- */</span></div> |
| <div class="line"><a name="l00012"></a><span class="lineno"> 12</span> </div> |
| <div class="line"><a name="l00013"></a><span class="lineno"> 13</span> m4_include(`SQLCommon.m4<span class="stringliteral">')</span></div> |
| <div class="line"><a name="l00014"></a><span class="lineno"> 14</span> <span class="stringliteral"></span><span class="comment"></span></div> |
| <div class="line"><a name="l00015"></a><span class="lineno"> 15</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00016"></a><span class="lineno"> 16</span> <span class="comment">@addtogroup grp_bayes</span></div> |
| <div class="line"><a name="l00017"></a><span class="lineno"> 17</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00018"></a><span class="lineno"> 18</span> <span class="comment">\warning <em> This MADlib method is still in early stage development. There may be some </span></div> |
| <div class="line"><a name="l00019"></a><span class="lineno"> 19</span> <span class="comment">issues that will be addressed in a future version. Interface and implementation</span></div> |
| <div class="line"><a name="l00020"></a><span class="lineno"> 20</span> <span class="comment">is subject to change. </em></span></div> |
| <div class="line"><a name="l00021"></a><span class="lineno"> 21</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00022"></a><span class="lineno"> 22</span> <span class="comment">@about</span></div> |
| <div class="line"><a name="l00023"></a><span class="lineno"> 23</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00024"></a><span class="lineno"> 24</span> <span class="comment">Naive Bayes refers to a stochastic model where all independent variables</span></div> |
| <div class="line"><a name="l00025"></a><span class="lineno"> 25</span> <span class="comment">\f$ a_1, \dots, a_n \f$ (often referred to as attributes in this context)</span></div> |
| <div class="line"><a name="l00026"></a><span class="lineno"> 26</span> <span class="comment">independently contribute to the probability that a data point belongs to a</span></div> |
| <div class="line"><a name="l00027"></a><span class="lineno"> 27</span> <span class="comment">certain class \f$ c \f$. In detail, \b Bayes' theorem states that</span></div> |
| <div class="line"><a name="l00028"></a><span class="lineno"> 28</span> <span class="comment">\f[</span></div> |
| <div class="line"><a name="l00029"></a><span class="lineno"> 29</span> <span class="comment"> \Pr(C = c \mid A_1 = a_1, \dots, A_n = a_n)</span></div> |
| <div class="line"><a name="l00030"></a><span class="lineno"> 30</span> <span class="comment"> = \frac{\Pr(C = c) \cdot \Pr(A_1 = a_1, \dots, A_n = a_n \mid C = c)}</span></div> |
| <div class="line"><a name="l00031"></a><span class="lineno"> 31</span> <span class="comment"> {\Pr(A_1 = a_1, \dots, A_n = a_n)}</span></div> |
| <div class="line"><a name="l00032"></a><span class="lineno"> 32</span> <span class="comment"> \,,</span></div> |
| <div class="line"><a name="l00033"></a><span class="lineno"> 33</span> <span class="comment">\f]</span></div> |
| <div class="line"><a name="l00034"></a><span class="lineno"> 34</span> <span class="comment">and the \b naive assumption is that</span></div> |
| <div class="line"><a name="l00035"></a><span class="lineno"> 35</span> <span class="comment">\f[</span></div> |
| <div class="line"><a name="l00036"></a><span class="lineno"> 36</span> <span class="comment"> \Pr(A_1 = a_1, \dots, A_n = a_n \mid C = c)</span></div> |
| <div class="line"><a name="l00037"></a><span class="lineno"> 37</span> <span class="comment"> = \prod_{i=1}^n \Pr(A_i = a_i \mid C = c)</span></div> |
| <div class="line"><a name="l00038"></a><span class="lineno"> 38</span> <span class="comment"> \,.</span></div> |
| <div class="line"><a name="l00039"></a><span class="lineno"> 39</span> <span class="comment">\f]</span></div> |
| <div class="line"><a name="l00040"></a><span class="lineno"> 40</span> <span class="comment">Naives Bayes classification estimates feature probabilities and class priors</span></div> |
| <div class="line"><a name="l00041"></a><span class="lineno"> 41</span> <span class="comment">using maximum likelihood or Laplacian smoothing. These parameters are then used</span></div> |
| <div class="line"><a name="l00042"></a><span class="lineno"> 42</span> <span class="comment">to classifying new data.</span></div> |
| <div class="line"><a name="l00043"></a><span class="lineno"> 43</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00044"></a><span class="lineno"> 44</span> <span class="comment">A Naive Bayes classifier computes the following formula:</span></div> |
| <div class="line"><a name="l00045"></a><span class="lineno"> 45</span> <span class="comment">\f[</span></div> |
| <div class="line"><a name="l00046"></a><span class="lineno"> 46</span> <span class="comment"> \text{classify}(a_1, ..., a_n)</span></div> |
| <div class="line"><a name="l00047"></a><span class="lineno"> 47</span> <span class="comment"> = \arg\max_c \left\{</span></div> |
| <div class="line"><a name="l00048"></a><span class="lineno"> 48</span> <span class="comment"> \Pr(C = c) \cdot \prod_{i=1}^n \Pr(A_i = a_i \mid C = c)</span></div> |
| <div class="line"><a name="l00049"></a><span class="lineno"> 49</span> <span class="comment"> \right\}</span></div> |
| <div class="line"><a name="l00050"></a><span class="lineno"> 50</span> <span class="comment">\f]</span></div> |
| <div class="line"><a name="l00051"></a><span class="lineno"> 51</span> <span class="comment">where \f$ c \f$ ranges over all classes in the training data and probabilites</span></div> |
| <div class="line"><a name="l00052"></a><span class="lineno"> 52</span> <span class="comment">are estimated with relative frequencies from the training set.</span></div> |
| <div class="line"><a name="l00053"></a><span class="lineno"> 53</span> <span class="comment">There are different ways to estimate the feature probabilities</span></div> |
| <div class="line"><a name="l00054"></a><span class="lineno"> 54</span> <span class="comment">\f$ P(A_i = a \mid C = c) \f$. The maximum likelihood estimate takes the</span></div> |
| <div class="line"><a name="l00055"></a><span class="lineno"> 55</span> <span class="comment">relative frequencies. That is:</span></div> |
| <div class="line"><a name="l00056"></a><span class="lineno"> 56</span> <span class="comment">\f[</span></div> |
| <div class="line"><a name="l00057"></a><span class="lineno"> 57</span> <span class="comment"> P(A_i = a \mid C = c) = \frac{\#(c,i,a)}{\#c}</span></div> |
| <div class="line"><a name="l00058"></a><span class="lineno"> 58</span> <span class="comment">\f]</span></div> |
| <div class="line"><a name="l00059"></a><span class="lineno"> 59</span> <span class="comment">where</span></div> |
| <div class="line"><a name="l00060"></a><span class="lineno"> 60</span> <span class="comment">- \f$ \#(c,i,a) \f$ denotes the # of training samples where attribute \f$ i \f$</span></div> |
| <div class="line"><a name="l00061"></a><span class="lineno"> 61</span> <span class="comment"> is \f$ a \f$ and class is \f$ c \f$</span></div> |
| <div class="line"><a name="l00062"></a><span class="lineno"> 62</span> <span class="comment">- \f$ \#c \f$ denotes the # of training samples where class is \f$ c \f$.</span></div> |
| <div class="line"><a name="l00063"></a><span class="lineno"> 63</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00064"></a><span class="lineno"> 64</span> <span class="comment">Since the maximum likelihood sometimes results in estimates of "0", you might</span></div> |
| <div class="line"><a name="l00065"></a><span class="lineno"> 65</span> <span class="comment">want to use a "smoothed" estimate. To do this, you add a number of "virtual"</span></div> |
| <div class="line"><a name="l00066"></a><span class="lineno"> 66</span> <span class="comment">samples and make the assumption that these samples are evenly distributed among</span></div> |
| <div class="line"><a name="l00067"></a><span class="lineno"> 67</span> <span class="comment">the values assumed by attribute \f$ i \f$ (that is, the set of all values</span></div> |
| <div class="line"><a name="l00068"></a><span class="lineno"> 68</span> <span class="comment">observed for attribute \f$ a \f$ for any class):</span></div> |
| <div class="line"><a name="l00069"></a><span class="lineno"> 69</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00070"></a><span class="lineno"> 70</span> <span class="comment">\f[</span></div> |
| <div class="line"><a name="l00071"></a><span class="lineno"> 71</span> <span class="comment"> P(A_i = a \mid C = c) = \frac{\#(c,i,a) + s}{\#c + s \cdot \#i}</span></div> |
| <div class="line"><a name="l00072"></a><span class="lineno"> 72</span> <span class="comment">\f]</span></div> |
| <div class="line"><a name="l00073"></a><span class="lineno"> 73</span> <span class="comment">where</span></div> |
| <div class="line"><a name="l00074"></a><span class="lineno"> 74</span> <span class="comment">- \f$ \#i \f$ denotes the # of distinct values for attribute \f$ i \f$ (for all</span></div> |
| <div class="line"><a name="l00075"></a><span class="lineno"> 75</span> <span class="comment"> classes)</span></div> |
| <div class="line"><a name="l00076"></a><span class="lineno"> 76</span> <span class="comment">- \f$ s \geq 0 \f$ denotes the smoothing factor.</span></div> |
| <div class="line"><a name="l00077"></a><span class="lineno"> 77</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00078"></a><span class="lineno"> 78</span> <span class="comment">The case \f$ s = 1 \f$ is known as "Laplace smoothing". The case \f$ s = 0 \f$</span></div> |
| <div class="line"><a name="l00079"></a><span class="lineno"> 79</span> <span class="comment">trivially reduces to maximum-likelihood estimates.</span></div> |
| <div class="line"><a name="l00080"></a><span class="lineno"> 80</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00081"></a><span class="lineno"> 81</span> <span class="comment">\b Note:</span></div> |
| <div class="line"><a name="l00082"></a><span class="lineno"> 82</span> <span class="comment">(1) The probabilities computed on the platforms of PostgreSQL and Greenplum</span></div> |
| <div class="line"><a name="l00083"></a><span class="lineno"> 83</span> <span class="comment">database have a small difference due to the nature of floating point</span></div> |
| <div class="line"><a name="l00084"></a><span class="lineno"> 84</span> <span class="comment">computation. Usually this is not important. However, if a data point has</span></div> |
| <div class="line"><a name="l00085"></a><span class="lineno"> 85</span> <span class="comment">\f[</span></div> |
| <div class="line"><a name="l00086"></a><span class="lineno"> 86</span> <span class="comment">P(C=c_i \mid A) \approx P(C=c_j \mid A)</span></div> |
| <div class="line"><a name="l00087"></a><span class="lineno"> 87</span> <span class="comment">\f]</span></div> |
| <div class="line"><a name="l00088"></a><span class="lineno"> 88</span> <span class="comment">for two classes, this data point might be classified into diferent classes on</span></div> |
| <div class="line"><a name="l00089"></a><span class="lineno"> 89</span> <span class="comment">PostgreSQL and Greenplum. This leads to the differences in classifications</span></div> |
| <div class="line"><a name="l00090"></a><span class="lineno"> 90</span> <span class="comment">on PostgreSQL and Greenplum for some data sets, but this should not</span></div> |
| <div class="line"><a name="l00091"></a><span class="lineno"> 91</span> <span class="comment">affect the quality of the results.</span></div> |
| <div class="line"><a name="l00092"></a><span class="lineno"> 92</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00093"></a><span class="lineno"> 93</span> <span class="comment">(2) When two classes have equal and highest probability among all classes,</span></div> |
| <div class="line"><a name="l00094"></a><span class="lineno"> 94</span> <span class="comment">the classification result is an array of these two classes, but the order</span></div> |
| <div class="line"><a name="l00095"></a><span class="lineno"> 95</span> <span class="comment">of the two classes is random.</span></div> |
| <div class="line"><a name="l00096"></a><span class="lineno"> 96</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00097"></a><span class="lineno"> 97</span> <span class="comment">(3) The current implementation of Naive Bayes classification is only suitable</span></div> |
| <div class="line"><a name="l00098"></a><span class="lineno"> 98</span> <span class="comment">for discontinuous (categorial) attributes.</span></div> |
| <div class="line"><a name="l00099"></a><span class="lineno"> 99</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00100"></a><span class="lineno"> 100</span> <span class="comment">For continuous data, a typical assumption, usually used for small datasets,</span></div> |
| <div class="line"><a name="l00101"></a><span class="lineno"> 101</span> <span class="comment">is that the continuous values associated with each class are distributed</span></div> |
| <div class="line"><a name="l00102"></a><span class="lineno"> 102</span> <span class="comment">according to a Gaussian distribution,</span></div> |
| <div class="line"><a name="l00103"></a><span class="lineno"> 103</span> <span class="comment">and then the probabilities \f$ P(A_i = a \mid C=c) \f$ can be estimated.</span></div> |
| <div class="line"><a name="l00104"></a><span class="lineno"> 104</span> <span class="comment">Another common technique for handling continuous values, which is better for</span></div> |
| <div class="line"><a name="l00105"></a><span class="lineno"> 105</span> <span class="comment">large data sets, is to use binning to discretize the values, and convert the</span></div> |
| <div class="line"><a name="l00106"></a><span class="lineno"> 106</span> <span class="comment">continuous data into categorical bins. These approaches are currently not</span></div> |
| <div class="line"><a name="l00107"></a><span class="lineno"> 107</span> <span class="comment">implemented and planned for future releases.</span></div> |
| <div class="line"><a name="l00108"></a><span class="lineno"> 108</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00109"></a><span class="lineno"> 109</span> <span class="comment">(4) One can still provide floating point data to the naive Bayes</span></div> |
| <div class="line"><a name="l00110"></a><span class="lineno"> 110</span> <span class="comment">classification function. Floating point numbers can be used as symbolic</span></div> |
| <div class="line"><a name="l00111"></a><span class="lineno"> 111</span> <span class="comment">substitutions for categorial data. The classification would work best if</span></div> |
| <div class="line"><a name="l00112"></a><span class="lineno"> 112</span> <span class="comment">there are sufficient data points for each floating point attribute. However,</span></div> |
| <div class="line"><a name="l00113"></a><span class="lineno"> 113</span> <span class="comment">if floating point numbers are used as continuous data, no warning is raised and</span></div> |
| <div class="line"><a name="l00114"></a><span class="lineno"> 114</span> <span class="comment">the result may not be as expected.</span></div> |
| <div class="line"><a name="l00115"></a><span class="lineno"> 115</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00116"></a><span class="lineno"> 116</span> <span class="comment">@input</span></div> |
| <div class="line"><a name="l00117"></a><span class="lineno"> 117</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00118"></a><span class="lineno"> 118</span> <span class="comment">The <b>training data</b> is expected to be of the following form:</span></div> |
| <div class="line"><a name="l00119"></a><span class="lineno"> 119</span> <span class="comment"><pre>{TABLE|VIEW} <em>trainingSource</em> (</span></div> |
| <div class="line"><a name="l00120"></a><span class="lineno"> 120</span> <span class="comment"> ...</span></div> |
| <div class="line"><a name="l00121"></a><span class="lineno"> 121</span> <span class="comment"> <em>trainingClassColumn</em> INTEGER,</span></div> |
| <div class="line"><a name="l00122"></a><span class="lineno"> 122</span> <span class="comment"> <em>trainingAttrColumn</em> INTEGER[],</span></div> |
| <div class="line"><a name="l00123"></a><span class="lineno"> 123</span> <span class="comment"> ...</span></div> |
| <div class="line"><a name="l00124"></a><span class="lineno"> 124</span> <span class="comment">)</pre></span></div> |
| <div class="line"><a name="l00125"></a><span class="lineno"> 125</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00126"></a><span class="lineno"> 126</span> <span class="comment">The <b>data to classify</b> is expected to be of the following form:</span></div> |
| <div class="line"><a name="l00127"></a><span class="lineno"> 127</span> <span class="comment"><pre>{TABLE|VIEW} <em>classifySource</em> (</span></div> |
| <div class="line"><a name="l00128"></a><span class="lineno"> 128</span> <span class="comment"> ...</span></div> |
| <div class="line"><a name="l00129"></a><span class="lineno"> 129</span> <span class="comment"> <em>classifyKeyColumn</em> ANYTYPE,</span></div> |
| <div class="line"><a name="l00130"></a><span class="lineno"> 130</span> <span class="comment"> <em>classifyAttrColumn</em> INTEGER[],</span></div> |
| <div class="line"><a name="l00131"></a><span class="lineno"> 131</span> <span class="comment"> ...</span></div> |
| <div class="line"><a name="l00132"></a><span class="lineno"> 132</span> <span class="comment">)</pre></span></div> |
| <div class="line"><a name="l00133"></a><span class="lineno"> 133</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00134"></a><span class="lineno"> 134</span> <span class="comment">@usage</span></div> |
| <div class="line"><a name="l00135"></a><span class="lineno"> 135</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00136"></a><span class="lineno"> 136</span> <span class="comment">- Precompute feature probabilities and class priors:</span></div> |
| <div class="line"><a name="l00137"></a><span class="lineno"> 137</span> <span class="comment"> <pre>SELECT \ref create_nb_prepared_data_tables(</span></div> |
| <div class="line"><a name="l00138"></a><span class="lineno"> 138</span> <span class="comment"> '<em>trainingSource</em>', '<em>trainingClassColumn</em>', '<em>trainingAttrColumn</em>',</span></div> |
| <div class="line"><a name="l00139"></a><span class="lineno"> 139</span> <span class="comment"> <em>numAttrs</em>, '<em>featureProbsName</em>', '<em>classPriorsName</em>'</span></div> |
| <div class="line"><a name="l00140"></a><span class="lineno"> 140</span> <span class="comment"> );</pre></span></div> |
| <div class="line"><a name="l00141"></a><span class="lineno"> 141</span> <span class="comment"> This creates table <em>featureProbsName</em> for storing feature</span></div> |
| <div class="line"><a name="l00142"></a><span class="lineno"> 142</span> <span class="comment"> probabilities and table <em>classPriorsName</em> for storing the class priors.</span></div> |
| <div class="line"><a name="l00143"></a><span class="lineno"> 143</span> <span class="comment">- Perform Naive Bayes classification:</span></div> |
| <div class="line"><a name="l00144"></a><span class="lineno"> 144</span> <span class="comment"> <pre>SELECT \ref create_nb_classify_view(</span></div> |
| <div class="line"><a name="l00145"></a><span class="lineno"> 145</span> <span class="comment"> '<em>featureProbsName</em>', '<em>classPriorsName</em>',</span></div> |
| <div class="line"><a name="l00146"></a><span class="lineno"> 146</span> <span class="comment"> '<em>classifySource</em>', '<em>classifyKeyColumn</em>', '<em>classifyAttrColumn</em>',</span></div> |
| <div class="line"><a name="l00147"></a><span class="lineno"> 147</span> <span class="comment"> <em>numAttrs</em>, '<em>destName</em>'</span></div> |
| <div class="line"><a name="l00148"></a><span class="lineno"> 148</span> <span class="comment"> );</pre></span></div> |
| <div class="line"><a name="l00149"></a><span class="lineno"> 149</span> <span class="comment"> This creates the view <tt><em>destName</em></tt> mapping</span></div> |
| <div class="line"><a name="l00150"></a><span class="lineno"> 150</span> <span class="comment"> <em>classifyKeyColumn</em> to the Naive Bayes classification:</span></div> |
| <div class="line"><a name="l00151"></a><span class="lineno"> 151</span> <span class="comment"> <pre>key | nb_classification</span></div> |
| <div class="line"><a name="l00152"></a><span class="lineno"> 152</span> <span class="comment">----+------------------</span></div> |
| <div class="line"><a name="l00153"></a><span class="lineno"> 153</span> <span class="comment">...</pre></span></div> |
| <div class="line"><a name="l00154"></a><span class="lineno"> 154</span> <span class="comment">- Compute Naive Bayes probabilities:</span></div> |
| <div class="line"><a name="l00155"></a><span class="lineno"> 155</span> <span class="comment"> <pre>SELECT \ref create_nb_probs_view(</span></div> |
| <div class="line"><a name="l00156"></a><span class="lineno"> 156</span> <span class="comment"> '<em>featureProbsName</em>', '<em>classPriorsName</em>',</span></div> |
| <div class="line"><a name="l00157"></a><span class="lineno"> 157</span> <span class="comment"> '<em>classifySource</em>', '<em>classifyKeyColumn</em>', '<em>classifyAttrColumn</em>',</span></div> |
| <div class="line"><a name="l00158"></a><span class="lineno"> 158</span> <span class="comment"> <em>numAttrs</em>, '<em>destName</em>'</span></div> |
| <div class="line"><a name="l00159"></a><span class="lineno"> 159</span> <span class="comment">);</pre></span></div> |
| <div class="line"><a name="l00160"></a><span class="lineno"> 160</span> <span class="comment"> This creates the view <tt><em>destName</em></tt> mapping</span></div> |
| <div class="line"><a name="l00161"></a><span class="lineno"> 161</span> <span class="comment"> <em>classifyKeyColumn</em> and every single class to the Naive Bayes</span></div> |
| <div class="line"><a name="l00162"></a><span class="lineno"> 162</span> <span class="comment"> probability:</span></div> |
| <div class="line"><a name="l00163"></a><span class="lineno"> 163</span> <span class="comment"> <pre>key | class | nb_prob</span></div> |
| <div class="line"><a name="l00164"></a><span class="lineno"> 164</span> <span class="comment">----+-------+--------</span></div> |
| <div class="line"><a name="l00165"></a><span class="lineno"> 165</span> <span class="comment">...</pre></span></div> |
| <div class="line"><a name="l00166"></a><span class="lineno"> 166</span> <span class="comment">- Ad-hoc execution (no precomputation):</span></div> |
| <div class="line"><a name="l00167"></a><span class="lineno"> 167</span> <span class="comment"> Functions \ref create_nb_classify_view and</span></div> |
| <div class="line"><a name="l00168"></a><span class="lineno"> 168</span> <span class="comment"> \ref create_nb_probs_view can be used in an ad-hoc fashion without the above</span></div> |
| <div class="line"><a name="l00169"></a><span class="lineno"> 169</span> <span class="comment"> precomputation step. In this case, replace the function arguments</span></div> |
| <div class="line"><a name="l00170"></a><span class="lineno"> 170</span> <span class="comment"> <pre>'<em>featureProbsName</em>', '<em>classPriorsName</em>'</pre></span></div> |
| <div class="line"><a name="l00171"></a><span class="lineno"> 171</span> <span class="comment"> with</span></div> |
| <div class="line"><a name="l00172"></a><span class="lineno"> 172</span> <span class="comment"> <pre>'<em>trainingSource</em>', '<em>trainingClassColumn</em>', '<em>trainingAttrColumn</em>'</pre></span></div> |
| <div class="line"><a name="l00173"></a><span class="lineno"> 173</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00174"></a><span class="lineno"> 174</span> <span class="comment">@examp</span></div> |
| <div class="line"><a name="l00175"></a><span class="lineno"> 175</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00176"></a><span class="lineno"> 176</span> <span class="comment">The following is an extremely simplified example of the above option #1 which</span></div> |
| <div class="line"><a name="l00177"></a><span class="lineno"> 177</span> <span class="comment">can by verified by hand.</span></div> |
| <div class="line"><a name="l00178"></a><span class="lineno"> 178</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00179"></a><span class="lineno"> 179</span> <span class="comment">-# The training and the classification data:</span></div> |
| <div class="line"><a name="l00180"></a><span class="lineno"> 180</span> <span class="comment">\verbatim</span></div> |
| <div class="line"><a name="l00181"></a><span class="lineno"> 181</span> <span class="comment">sql> SELECT * FROM training;</span></div> |
| <div class="line"><a name="l00182"></a><span class="lineno"> 182</span> <span class="comment"> id | class | attributes</span></div> |
| <div class="line"><a name="l00183"></a><span class="lineno"> 183</span> <span class="comment">----+-------+------------</span></div> |
| <div class="line"><a name="l00184"></a><span class="lineno"> 184</span> <span class="comment"> 1 | 1 | {1,2,3}</span></div> |
| <div class="line"><a name="l00185"></a><span class="lineno"> 185</span> <span class="comment"> 2 | 1 | {1,2,1}</span></div> |
| <div class="line"><a name="l00186"></a><span class="lineno"> 186</span> <span class="comment"> 3 | 1 | {1,4,3}</span></div> |
| <div class="line"><a name="l00187"></a><span class="lineno"> 187</span> <span class="comment"> 4 | 2 | {1,2,2}</span></div> |
| <div class="line"><a name="l00188"></a><span class="lineno"> 188</span> <span class="comment"> 5 | 2 | {0,2,2}</span></div> |
| <div class="line"><a name="l00189"></a><span class="lineno"> 189</span> <span class="comment"> 6 | 2 | {0,1,3}</span></div> |
| <div class="line"><a name="l00190"></a><span class="lineno"> 190</span> <span class="comment">(6 rows)</span></div> |
| <div class="line"><a name="l00191"></a><span class="lineno"> 191</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00192"></a><span class="lineno"> 192</span> <span class="comment">sql> select * from toclassify;</span></div> |
| <div class="line"><a name="l00193"></a><span class="lineno"> 193</span> <span class="comment"> id | attributes</span></div> |
| <div class="line"><a name="l00194"></a><span class="lineno"> 194</span> <span class="comment">----+------------</span></div> |
| <div class="line"><a name="l00195"></a><span class="lineno"> 195</span> <span class="comment"> 1 | {0,2,1}</span></div> |
| <div class="line"><a name="l00196"></a><span class="lineno"> 196</span> <span class="comment"> 2 | {1,2,3}</span></div> |
| <div class="line"><a name="l00197"></a><span class="lineno"> 197</span> <span class="comment">(2 rows)</span></div> |
| <div class="line"><a name="l00198"></a><span class="lineno"> 198</span> <span class="comment">\endverbatim</span></div> |
| <div class="line"><a name="l00199"></a><span class="lineno"> 199</span> <span class="comment">-# Precompute feature probabilities and class priors</span></div> |
| <div class="line"><a name="l00200"></a><span class="lineno"> 200</span> <span class="comment">\verbatim</span></div> |
| <div class="line"><a name="l00201"></a><span class="lineno"> 201</span> <span class="comment">sql> SELECT madlib.create_nb_prepared_data_tables(</span></div> |
| <div class="line"><a name="l00202"></a><span class="lineno"> 202</span> <span class="comment">'training', 'class', 'attributes', 3, 'nb_feature_probs', 'nb_class_priors');</span></div> |
| <div class="line"><a name="l00203"></a><span class="lineno"> 203</span> <span class="comment">\endverbatim</span></div> |
| <div class="line"><a name="l00204"></a><span class="lineno"> 204</span> <span class="comment">-# Optionally check the contents of the precomputed tables:</span></div> |
| <div class="line"><a name="l00205"></a><span class="lineno"> 205</span> <span class="comment">\verbatim</span></div> |
| <div class="line"><a name="l00206"></a><span class="lineno"> 206</span> <span class="comment">sql> SELECT * FROM nb_class_priors;</span></div> |
| <div class="line"><a name="l00207"></a><span class="lineno"> 207</span> <span class="comment"> class | class_cnt | all_cnt</span></div> |
| <div class="line"><a name="l00208"></a><span class="lineno"> 208</span> <span class="comment">-------+-----------+---------</span></div> |
| <div class="line"><a name="l00209"></a><span class="lineno"> 209</span> <span class="comment"> 1 | 3 | 6</span></div> |
| <div class="line"><a name="l00210"></a><span class="lineno"> 210</span> <span class="comment"> 2 | 3 | 6</span></div> |
| <div class="line"><a name="l00211"></a><span class="lineno"> 211</span> <span class="comment">(2 rows)</span></div> |
| <div class="line"><a name="l00212"></a><span class="lineno"> 212</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00213"></a><span class="lineno"> 213</span> <span class="comment">sql> SELECT * FROM nb_feature_probs;</span></div> |
| <div class="line"><a name="l00214"></a><span class="lineno"> 214</span> <span class="comment"> class | attr | value | cnt | attr_cnt</span></div> |
| <div class="line"><a name="l00215"></a><span class="lineno"> 215</span> <span class="comment">-------+------+-------+-----+----------</span></div> |
| <div class="line"><a name="l00216"></a><span class="lineno"> 216</span> <span class="comment"> 1 | 1 | 0 | 0 | 2</span></div> |
| <div class="line"><a name="l00217"></a><span class="lineno"> 217</span> <span class="comment"> 1 | 1 | 1 | 3 | 2</span></div> |
| <div class="line"><a name="l00218"></a><span class="lineno"> 218</span> <span class="comment"> 1 | 2 | 1 | 0 | 3</span></div> |
| <div class="line"><a name="l00219"></a><span class="lineno"> 219</span> <span class="comment"> 1 | 2 | 2 | 2 | 3</span></div> |
| <div class="line"><a name="l00220"></a><span class="lineno"> 220</span> <span class="comment">...</span></div> |
| <div class="line"><a name="l00221"></a><span class="lineno"> 221</span> <span class="comment">\endverbatim</span></div> |
| <div class="line"><a name="l00222"></a><span class="lineno"> 222</span> <span class="comment">-# Create the view with Naive Bayes classification and check the results:</span></div> |
| <div class="line"><a name="l00223"></a><span class="lineno"> 223</span> <span class="comment">\verbatim</span></div> |
| <div class="line"><a name="l00224"></a><span class="lineno"> 224</span> <span class="comment">sql> SELECT madlib.create_nb_classify_view (</span></div> |
| <div class="line"><a name="l00225"></a><span class="lineno"> 225</span> <span class="comment">'nb_feature_probs', 'nb_class_priors', 'toclassify', 'id', 'attributes', 3, 'nb_classify_view_fast');</span></div> |
| <div class="line"><a name="l00226"></a><span class="lineno"> 226</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00227"></a><span class="lineno"> 227</span> <span class="comment">sql> SELECT * FROM nb_classify_view_fast;</span></div> |
| <div class="line"><a name="l00228"></a><span class="lineno"> 228</span> <span class="comment"> key | nb_classification</span></div> |
| <div class="line"><a name="l00229"></a><span class="lineno"> 229</span> <span class="comment">-----+-------------------</span></div> |
| <div class="line"><a name="l00230"></a><span class="lineno"> 230</span> <span class="comment"> 1 | {2}</span></div> |
| <div class="line"><a name="l00231"></a><span class="lineno"> 231</span> <span class="comment"> 2 | {1}</span></div> |
| <div class="line"><a name="l00232"></a><span class="lineno"> 232</span> <span class="comment">(2 rows)</span></div> |
| <div class="line"><a name="l00233"></a><span class="lineno"> 233</span> <span class="comment">\endverbatim</span></div> |
| <div class="line"><a name="l00234"></a><span class="lineno"> 234</span> <span class="comment">-# Look at the probabilities for each class (note that we use "Laplacian smoothing"):</span></div> |
| <div class="line"><a name="l00235"></a><span class="lineno"> 235</span> <span class="comment">\verbatim</span></div> |
| <div class="line"><a name="l00236"></a><span class="lineno"> 236</span> <span class="comment">sql> SELECT madlib.create_nb_probs_view (</span></div> |
| <div class="line"><a name="l00237"></a><span class="lineno"> 237</span> <span class="comment">'nb_feature_probs', 'nb_class_priors', 'toclassify', 'id', 'attributes', 3, 'nb_probs_view_fast');</span></div> |
| <div class="line"><a name="l00238"></a><span class="lineno"> 238</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00239"></a><span class="lineno"> 239</span> <span class="comment">sql> SELECT * FROM nb_probs_view_fast;</span></div> |
| <div class="line"><a name="l00240"></a><span class="lineno"> 240</span> <span class="comment"> key | class | nb_prob</span></div> |
| <div class="line"><a name="l00241"></a><span class="lineno"> 241</span> <span class="comment">-----+-------+---------</span></div> |
| <div class="line"><a name="l00242"></a><span class="lineno"> 242</span> <span class="comment"> 1 | 1 | 0.4</span></div> |
| <div class="line"><a name="l00243"></a><span class="lineno"> 243</span> <span class="comment"> 1 | 2 | 0.6</span></div> |
| <div class="line"><a name="l00244"></a><span class="lineno"> 244</span> <span class="comment"> 2 | 1 | 0.75</span></div> |
| <div class="line"><a name="l00245"></a><span class="lineno"> 245</span> <span class="comment"> 2 | 2 | 0.25</span></div> |
| <div class="line"><a name="l00246"></a><span class="lineno"> 246</span> <span class="comment">(4 rows)</span></div> |
| <div class="line"><a name="l00247"></a><span class="lineno"> 247</span> <span class="comment">\endverbatim</span></div> |
| <div class="line"><a name="l00248"></a><span class="lineno"> 248</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00249"></a><span class="lineno"> 249</span> <span class="comment">@literature</span></div> |
| <div class="line"><a name="l00250"></a><span class="lineno"> 250</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00251"></a><span class="lineno"> 251</span> <span class="comment">[1] Tom Mitchell: Machine Learning, McGraw Hill, 1997. Book chapter</span></div> |
| <div class="line"><a name="l00252"></a><span class="lineno"> 252</span> <span class="comment"> <em>Generativ and Discriminative Classifiers: Naive Bayes and Logistic</span></div> |
| <div class="line"><a name="l00253"></a><span class="lineno"> 253</span> <span class="comment"> Regression</em> available at: http://www.cs.cmu.edu/~tom/NewChapters.html</span></div> |
| <div class="line"><a name="l00254"></a><span class="lineno"> 254</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00255"></a><span class="lineno"> 255</span> <span class="comment">[2] Wikipedia, Naive Bayes classifier,</span></div> |
| <div class="line"><a name="l00256"></a><span class="lineno"> 256</span> <span class="comment"> http://en.wikipedia.org/wiki/Naive_Bayes_classifier</span></div> |
| <div class="line"><a name="l00257"></a><span class="lineno"> 257</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00258"></a><span class="lineno"> 258</span> <span class="comment">@sa File bayes.sql_in documenting the SQL functions.</span></div> |
| <div class="line"><a name="l00259"></a><span class="lineno"> 259</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00260"></a><span class="lineno"> 260</span> <span class="comment">@internal</span></div> |
| <div class="line"><a name="l00261"></a><span class="lineno"> 261</span> <span class="comment">@sa namespace bayes (documenting the implementation in Python)</span></div> |
| <div class="line"><a name="l00262"></a><span class="lineno"> 262</span> <span class="comment">@endinternal</span></div> |
| <div class="line"><a name="l00263"></a><span class="lineno"> 263</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00264"></a><span class="lineno"> 264</span> <span class="comment">*/</span></div> |
| <div class="line"><a name="l00265"></a><span class="lineno"> 265</span> </div> |
| <div class="line"><a name="l00266"></a><span class="lineno"> 266</span> -- Begin of argmax definition</div> |
| <div class="line"><a name="l00267"></a><span class="lineno"> 267</span> </div> |
| <div class="line"><a name="l00268"></a><span class="lineno"> 268</span> CREATE TYPE MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE AS (</div> |
| <div class="line"><a name="l00269"></a><span class="lineno"> 269</span>  args INTEGER[],</div> |
| <div class="line"><a name="l00270"></a><span class="lineno"> 270</span>  value DOUBLE PRECISION</div> |
| <div class="line"><a name="l00271"></a><span class="lineno"> 271</span> );</div> |
| <div class="line"><a name="l00272"></a><span class="lineno"> 272</span> </div> |
| <div class="line"><a name="l00273"></a><span class="lineno"> 273</span> CREATE FUNCTION MADLIB_SCHEMA.argmax_transition(</div> |
| <div class="line"><a name="l00274"></a><span class="lineno"> 274</span>  oldmax MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE,</div> |
| <div class="line"><a name="l00275"></a><span class="lineno"> 275</span>  newkey INTEGER,</div> |
| <div class="line"><a name="l00276"></a><span class="lineno"> 276</span>  newvalue DOUBLE PRECISION)</div> |
| <div class="line"><a name="l00277"></a><span class="lineno"> 277</span> RETURNS MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE AS</div> |
| <div class="line"><a name="l00278"></a><span class="lineno"> 278</span> $$</div> |
| <div class="line"><a name="l00279"></a><span class="lineno"> 279</span>  SELECT CASE WHEN $3 < $1.value OR $2 IS NULL OR ($3 IS NULL AND NOT $1.value IS NULL) THEN $1</div> |
| <div class="line"><a name="l00280"></a><span class="lineno"> 280</span>  WHEN $3 = $1.value OR ($3 IS NULL AND $1.value IS NULL AND NOT $1.args IS NULL)</div> |
| <div class="line"><a name="l00281"></a><span class="lineno"> 281</span>  THEN ($1.args || $2, $3)::MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE</div> |
| <div class="line"><a name="l00282"></a><span class="lineno"> 282</span>  ELSE (array[$2], $3)::MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE</div> |
| <div class="line"><a name="l00283"></a><span class="lineno"> 283</span>  END</div> |
| <div class="line"><a name="l00284"></a><span class="lineno"> 284</span> $$</div> |
| <div class="line"><a name="l00285"></a><span class="lineno"> 285</span> LANGUAGE sql IMMUTABLE;</div> |
| <div class="line"><a name="l00286"></a><span class="lineno"> 286</span> </div> |
| <div class="line"><a name="l00287"></a><span class="lineno"> 287</span> CREATE FUNCTION MADLIB_SCHEMA.argmax_combine(</div> |
| <div class="line"><a name="l00288"></a><span class="lineno"> 288</span>  max1 MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE,</div> |
| <div class="line"><a name="l00289"></a><span class="lineno"> 289</span>  max2 MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE)</div> |
| <div class="line"><a name="l00290"></a><span class="lineno"> 290</span> RETURNS MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE AS</div> |
| <div class="line"><a name="l00291"></a><span class="lineno"> 291</span> $$</div> |
| <div class="line"><a name="l00292"></a><span class="lineno"> 292</span>  -- If SQL guaranteed short-circuit evaluation, the following could become</div> |
| <div class="line"><a name="l00293"></a><span class="lineno"> 293</span>  -- shorter. Unfortunately, this is not the case.</div> |
| <div class="line"><a name="l00294"></a><span class="lineno"> 294</span>  -- Section 6.3.3.3 of ISO/IEC 9075-1:2008 Framework (SQL/Framework):</div> |
| <div class="line"><a name="l00295"></a><span class="lineno"> 295</span>  --</div> |
| <div class="line"><a name="l00296"></a><span class="lineno"> 296</span>  -- "However, it is implementation-dependent whether expressions are</div> |
| <div class="line"><a name="l00297"></a><span class="lineno"> 297</span>  -- actually evaluated left to right, particularly when operands or</div> |
| <div class="line"><a name="l00298"></a><span class="lineno"> 298</span>  -- operators might cause conditions to be raised or if the results of the</div> |
| <div class="line"><a name="l00299"></a><span class="lineno"> 299</span>  -- expressions can be determined without completely evaluating all parts</div> |
| <div class="line"><a name="l00300"></a><span class="lineno"> 300</span>  -- of the expression."</div> |
| <div class="line"><a name="l00301"></a><span class="lineno"> 301</span>  --</div> |
| <div class="line"><a name="l00302"></a><span class="lineno"> 302</span>  -- Again, the optimizer does its job hopefully.</div> |
| <div class="line"><a name="l00303"></a><span class="lineno"> 303</span>  SELECT CASE WHEN $1 IS NULL THEN $2</div> |
| <div class="line"><a name="l00304"></a><span class="lineno"> 304</span>  WHEN $2 IS NULL THEN $1</div> |
| <div class="line"><a name="l00305"></a><span class="lineno"> 305</span>  WHEN ($1.value = $2.value) OR ($1.value IS NULL AND $2.value IS NULL)</div> |
| <div class="line"><a name="l00306"></a><span class="lineno"> 306</span>  THEN ($1.args || $2.args, $1.value)::MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE</div> |
| <div class="line"><a name="l00307"></a><span class="lineno"> 307</span>  WHEN $1.value IS NULL OR $1.value < $2.value THEN $2</div> |
| <div class="line"><a name="l00308"></a><span class="lineno"> 308</span>  ELSE $1</div> |
| <div class="line"><a name="l00309"></a><span class="lineno"> 309</span>  END</div> |
| <div class="line"><a name="l00310"></a><span class="lineno"> 310</span> $$</div> |
| <div class="line"><a name="l00311"></a><span class="lineno"> 311</span> LANGUAGE sql IMMUTABLE;</div> |
| <div class="line"><a name="l00312"></a><span class="lineno"> 312</span> </div> |
| <div class="line"><a name="l00313"></a><span class="lineno"> 313</span> CREATE FUNCTION MADLIB_SCHEMA.argmax_final(</div> |
| <div class="line"><a name="l00314"></a><span class="lineno"> 314</span>  finalstate MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE)</div> |
| <div class="line"><a name="l00315"></a><span class="lineno"> 315</span> RETURNS INTEGER[] AS</div> |
| <div class="line"><a name="l00316"></a><span class="lineno"> 316</span> $$</div> |
| <div class="line"><a name="l00317"></a><span class="lineno"> 317</span>  SELECT $1.args</div> |
| <div class="line"><a name="l00318"></a><span class="lineno"> 318</span> $$</div> |
| <div class="line"><a name="l00319"></a><span class="lineno"> 319</span> LANGUAGE sql IMMUTABLE;</div> |
| <div class="line"><a name="l00320"></a><span class="lineno"> 320</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00321"></a><span class="lineno"> 321</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00322"></a><span class="lineno"> 322</span> <span class="comment"> * @internal</span></div> |
| <div class="line"><a name="l00323"></a><span class="lineno"> 323</span> <span class="comment"> * @brief Argmax: Return the key of the row for which value is maximal</span></div> |
| <div class="line"><a name="l00324"></a><span class="lineno"> 324</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00325"></a><span class="lineno"> 325</span> <span class="comment"> * The "index set" of the argmax function is of type INTEGER and we range over</span></div> |
| <div class="line"><a name="l00326"></a><span class="lineno"> 326</span> <span class="comment"> * DOUBLE PRECISION values. It is not required that all keys are distinct.</span></div> |
| <div class="line"><a name="l00327"></a><span class="lineno"> 327</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00328"></a><span class="lineno"> 328</span> <span class="comment"> * @note</span></div> |
| <div class="line"><a name="l00329"></a><span class="lineno"> 329</span> <span class="comment"> * argmax should only be used on unsorted data because it will not exploit</span></div> |
| <div class="line"><a name="l00330"></a><span class="lineno"> 330</span> <span class="comment"> * indices, and its running time is \f$ \Theta(n) \f$.</span></div> |
| <div class="line"><a name="l00331"></a><span class="lineno"> 331</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00332"></a><span class="lineno"> 332</span> <span class="comment"> * @implementation</span></div> |
| <div class="line"><a name="l00333"></a><span class="lineno"> 333</span> <span class="comment"> * The implementation is in SQL, with a flavor of functional programming.</span></div> |
| <div class="line"><a name="l00334"></a><span class="lineno"> 334</span> <span class="comment"> * The hope is that the optimizer does a good job here.</span></div> |
| <div class="line"><a name="l00335"></a><span class="lineno"> 335</span> <span class="comment"> */</span></div> |
| <div class="line"><a name="l00336"></a><span class="lineno"> 336</span> CREATE AGGREGATE MADLIB_SCHEMA.argmax(/*+ key */ INTEGER, /*+ value */ DOUBLE PRECISION) (</div> |
| <div class="line"><a name="l00337"></a><span class="lineno"> 337</span>  SFUNC=MADLIB_SCHEMA.argmax_transition,</div> |
| <div class="line"><a name="l00338"></a><span class="lineno"> 338</span>  STYPE=MADLIB_SCHEMA.ARGS_AND_VALUE_DOUBLE,</div> |
| <div class="line"><a name="l00339"></a><span class="lineno"> 339</span>  m4_ifdef(`__GREENPLUM__',`prefunc=MADLIB_SCHEMA.argmax_combine,<span class="stringliteral">')</span></div> |
| <div class="line"><a name="l00340"></a><span class="lineno"> 340</span> <span class="stringliteral"> FINALFUNC=MADLIB_SCHEMA.argmax_final</span></div> |
| <div class="line"><a name="l00341"></a><span class="lineno"> 341</span> <span class="stringliteral">);</span></div> |
| <div class="line"><a name="l00342"></a><span class="lineno"> 342</span> <span class="stringliteral"></span></div> |
| <div class="line"><a name="l00343"></a><span class="lineno"> 343</span> <span class="stringliteral"></span><span class="comment"></span></div> |
| <div class="line"><a name="l00344"></a><span class="lineno"> 344</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00345"></a><span class="lineno"> 345</span> <span class="comment"> * @brief Precompute all class priors and feature probabilities</span></div> |
| <div class="line"><a name="l00346"></a><span class="lineno"> 346</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00347"></a><span class="lineno"> 347</span> <span class="comment"> * Feature probabilities are stored in a table of format</span></div> |
| <div class="line"><a name="l00348"></a><span class="lineno"> 348</span> <span class="comment"> * <pre>TABLE <em>featureProbsDestName</em> (</span></div> |
| <div class="line"><a name="l00349"></a><span class="lineno"> 349</span> <span class="comment"> * class INTEGER,</span></div> |
| <div class="line"><a name="l00350"></a><span class="lineno"> 350</span> <span class="comment"> * attr INTEGER,</span></div> |
| <div class="line"><a name="l00351"></a><span class="lineno"> 351</span> <span class="comment"> * value INTEGER,</span></div> |
| <div class="line"><a name="l00352"></a><span class="lineno"> 352</span> <span class="comment"> * cnt INTEGER,</span></div> |
| <div class="line"><a name="l00353"></a><span class="lineno"> 353</span> <span class="comment"> * attr_cnt INTEGER</span></div> |
| <div class="line"><a name="l00354"></a><span class="lineno"> 354</span> <span class="comment"> *)</pre></span></div> |
| <div class="line"><a name="l00355"></a><span class="lineno"> 355</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00356"></a><span class="lineno"> 356</span> <span class="comment"> * Class priors are stored in a table of format</span></div> |
| <div class="line"><a name="l00357"></a><span class="lineno"> 357</span> <span class="comment"> * <pre>TABLE <em>classPriorsDestName</em> (</span></div> |
| <div class="line"><a name="l00358"></a><span class="lineno"> 358</span> <span class="comment"> * class INTEGER,</span></div> |
| <div class="line"><a name="l00359"></a><span class="lineno"> 359</span> <span class="comment"> * class_cnt INTEGER,</span></div> |
| <div class="line"><a name="l00360"></a><span class="lineno"> 360</span> <span class="comment"> * all_cnt INTEGER</span></div> |
| <div class="line"><a name="l00361"></a><span class="lineno"> 361</span> <span class="comment"> *)</pre></span></div> |
| <div class="line"><a name="l00362"></a><span class="lineno"> 362</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00363"></a><span class="lineno"> 363</span> <span class="comment"> * @param trainingSource Name of relation containing the training data</span></div> |
| <div class="line"><a name="l00364"></a><span class="lineno"> 364</span> <span class="comment"> * @param trainingClassColumn Name of class column in training data</span></div> |
| <div class="line"><a name="l00365"></a><span class="lineno"> 365</span> <span class="comment"> * @param trainingAttrColumn Name of attributes-array column in training data</span></div> |
| <div class="line"><a name="l00366"></a><span class="lineno"> 366</span> <span class="comment"> * @param numAttrs Number of attributes to use for classification</span></div> |
| <div class="line"><a name="l00367"></a><span class="lineno"> 367</span> <span class="comment"> * @param featureProbsDestName Name of feature-probabilities table to create</span></div> |
| <div class="line"><a name="l00368"></a><span class="lineno"> 368</span> <span class="comment"> * @param classPriorsDestName Name of class-priors table to create</span></div> |
| <div class="line"><a name="l00369"></a><span class="lineno"> 369</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00370"></a><span class="lineno"> 370</span> <span class="comment"> * @usage</span></div> |
| <div class="line"><a name="l00371"></a><span class="lineno"> 371</span> <span class="comment"> * Precompute feature probabilities and class priors:</span></div> |
| <div class="line"><a name="l00372"></a><span class="lineno"> 372</span> <span class="comment"> * <pre>SELECT \ref create_nb_prepared_data_tables(</span></div> |
| <div class="line"><a name="l00373"></a><span class="lineno"> 373</span> <span class="comment"> * '<em>trainingSource</em>', '<em>trainingClassColumn</em>', '<em>trainingAttrColumn</em>',</span></div> |
| <div class="line"><a name="l00374"></a><span class="lineno"> 374</span> <span class="comment"> * <em>numAttrs</em>, '<em>featureProbsName</em>', '<em>classPriorsName</em>'</span></div> |
| <div class="line"><a name="l00375"></a><span class="lineno"> 375</span> <span class="comment"> *);</pre></span></div> |
| <div class="line"><a name="l00376"></a><span class="lineno"> 376</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00377"></a><span class="lineno"> 377</span> <span class="comment"> * @internal</span></div> |
| <div class="line"><a name="l00378"></a><span class="lineno"> 378</span> <span class="comment"> * @sa This function is a wrapper for bayes::create_prepared_data().</span></div> |
| <div class="line"><a name="l00379"></a><span class="lineno"> 379</span> <span class="comment"> */</span></div> |
| <div class="line"><a name="l00380"></a><span class="lineno"> 380</span> CREATE FUNCTION MADLIB_SCHEMA.create_nb_prepared_data_tables(</div> |
| <div class="line"><a name="l00381"></a><span class="lineno"> 381</span>  "trainingSource" VARCHAR,</div> |
| <div class="line"><a name="l00382"></a><span class="lineno"> 382</span>  "trainingClassColumn" VARCHAR,</div> |
| <div class="line"><a name="l00383"></a><span class="lineno"> 383</span>  "trainingAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00384"></a><span class="lineno"> 384</span>  "numAttrs" INTEGER,</div> |
| <div class="line"><a name="l00385"></a><span class="lineno"> 385</span>  "featureProbsDestName" VARCHAR,</div> |
| <div class="line"><a name="l00386"></a><span class="lineno"> 386</span>  "classPriorsDestName" VARCHAR)</div> |
| <div class="line"><a name="l00387"></a><span class="lineno"> 387</span> RETURNS VOID</div> |
| <div class="line"><a name="l00388"></a><span class="lineno"> 388</span> AS $$PythonFunction(bayes, bayes, create_prepared_data_table)$$</div> |
| <div class="line"><a name="l00389"></a><span class="lineno"> 389</span> LANGUAGE plpythonu VOLATILE;</div> |
| <div class="line"><a name="l00390"></a><span class="lineno"> 390</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00391"></a><span class="lineno"> 391</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00392"></a><span class="lineno"> 392</span> <span class="comment"> * @brief Create a view with columns <tt>(key, nb_classification)</tt></span></div> |
| <div class="line"><a name="l00393"></a><span class="lineno"> 393</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00394"></a><span class="lineno"> 394</span> <span class="comment"> * The created relation will be</span></div> |
| <div class="line"><a name="l00395"></a><span class="lineno"> 395</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00396"></a><span class="lineno"> 396</span> <span class="comment"> * <tt>{TABLE|VIEW} <em>destName</em> (key, nb_classification)</tt></span></div> |
| <div class="line"><a name="l00397"></a><span class="lineno"> 397</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00398"></a><span class="lineno"> 398</span> <span class="comment"> * where \c nb_classification is an array containing the most likely</span></div> |
| <div class="line"><a name="l00399"></a><span class="lineno"> 399</span> <span class="comment"> * class(es) of the record in \em classifySource identified by \c key.</span></div> |
| <div class="line"><a name="l00400"></a><span class="lineno"> 400</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00401"></a><span class="lineno"> 401</span> <span class="comment"> * @param featureProbsSource Name of table with precomputed feature</span></div> |
| <div class="line"><a name="l00402"></a><span class="lineno"><a class="code" href="bayes_8sql__in.html#aeb4eae7843dd789cc38d5fc57f4ccfb2"> 402</a></span> <span class="comment"> * probabilities, as created with create_nb_prepared_data_tables()</span></div> |
| <div class="line"><a name="l00403"></a><span class="lineno"> 403</span> <span class="comment"> * @param classPriorsSource Name of table with precomputed class priors, as</span></div> |
| <div class="line"><a name="l00404"></a><span class="lineno"> 404</span> <span class="comment"> * created with create_nb_prepared_data_tables()</span></div> |
| <div class="line"><a name="l00405"></a><span class="lineno"> 405</span> <span class="comment"> * @param classifySource Name of the relation that contains data to be classified</span></div> |
| <div class="line"><a name="l00406"></a><span class="lineno"> 406</span> <span class="comment"> * @param classifyKeyColumn Name of column in \em classifySource that can</span></div> |
| <div class="line"><a name="l00407"></a><span class="lineno"> 407</span> <span class="comment"> * serve as unique identifier (the key of the source relation)</span></div> |
| <div class="line"><a name="l00408"></a><span class="lineno"> 408</span> <span class="comment"> * @param classifyAttrColumn Name of attributes-array column in \em classifySource</span></div> |
| <div class="line"><a name="l00409"></a><span class="lineno"> 409</span> <span class="comment"> * @param numAttrs Number of attributes to use for classification</span></div> |
| <div class="line"><a name="l00410"></a><span class="lineno"> 410</span> <span class="comment"> * @param destName Name of the view to create</span></div> |
| <div class="line"><a name="l00411"></a><span class="lineno"> 411</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00412"></a><span class="lineno"> 412</span> <span class="comment"> * @note \c create_nb_classify_view can be called in an ad-hoc fashion. See</span></div> |
| <div class="line"><a name="l00413"></a><span class="lineno"> 413</span> <span class="comment"> * \ref grp_bayes for instructions.</span></div> |
| <div class="line"><a name="l00414"></a><span class="lineno"> 414</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00415"></a><span class="lineno"> 415</span> <span class="comment"> * @usage</span></div> |
| <div class="line"><a name="l00416"></a><span class="lineno"> 416</span> <span class="comment"> * -# Create Naive Bayes classifications view:</span></div> |
| <div class="line"><a name="l00417"></a><span class="lineno"> 417</span> <span class="comment"> * <pre>SELECT \ref create_nb_classify_view(</span></div> |
| <div class="line"><a name="l00418"></a><span class="lineno"> 418</span> <span class="comment"> * '<em>featureProbsName</em>', '<em>classPriorsName</em>',</span></div> |
| <div class="line"><a name="l00419"></a><span class="lineno"> 419</span> <span class="comment"> * '<em>classifySource</em>', '<em>classifyKeyColumn</em>', '<em>classifyAttrColumn</em>',</span></div> |
| <div class="line"><a name="l00420"></a><span class="lineno"> 420</span> <span class="comment"> * <em>numAttrs</em>, '<em>destName</em>'</span></div> |
| <div class="line"><a name="l00421"></a><span class="lineno"> 421</span> <span class="comment"> *);</pre></span></div> |
| <div class="line"><a name="l00422"></a><span class="lineno"> 422</span> <span class="comment"> * -# Show Naive Bayes classifications:</span></div> |
| <div class="line"><a name="l00423"></a><span class="lineno"> 423</span> <span class="comment"> * <pre>SELECT * FROM <em>destName</em>;</pre></span></div> |
| <div class="line"><a name="l00424"></a><span class="lineno"> 424</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00425"></a><span class="lineno"> 425</span> <span class="comment"> * @internal</span></div> |
| <div class="line"><a name="l00426"></a><span class="lineno"> 426</span> <span class="comment"> * @sa This function is a wrapper for bayes::create_classification(). See there</span></div> |
| <div class="line"><a name="l00427"></a><span class="lineno"> 427</span> <span class="comment"> * for details.</span></div> |
| <div class="line"><a name="l00428"></a><span class="lineno"> 428</span> <span class="comment"> */</span></div> |
| <div class="line"><a name="l00429"></a><span class="lineno"> 429</span> CREATE FUNCTION MADLIB_SCHEMA.create_nb_classify_view(</div> |
| <div class="line"><a name="l00430"></a><span class="lineno"> 430</span>  "featureProbsSource" VARCHAR,</div> |
| <div class="line"><a name="l00431"></a><span class="lineno"> 431</span>  "classPriorsSource" VARCHAR,</div> |
| <div class="line"><a name="l00432"></a><span class="lineno"> 432</span>  "classifySource" VARCHAR,</div> |
| <div class="line"><a name="l00433"></a><span class="lineno"> 433</span>  "classifyKeyColumn" VARCHAR,</div> |
| <div class="line"><a name="l00434"></a><span class="lineno"> 434</span>  "classifyAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00435"></a><span class="lineno"> 435</span>  "numAttrs" INTEGER,</div> |
| <div class="line"><a name="l00436"></a><span class="lineno"> 436</span>  "destName" VARCHAR)</div> |
| <div class="line"><a name="l00437"></a><span class="lineno"> 437</span> RETURNS VOID</div> |
| <div class="line"><a name="l00438"></a><span class="lineno"> 438</span> AS $$PythonFunction(bayes, bayes, create_classification_view)$$</div> |
| <div class="line"><a name="l00439"></a><span class="lineno"> 439</span> LANGUAGE plpythonu VOLATILE;</div> |
| <div class="line"><a name="l00440"></a><span class="lineno"> 440</span> </div> |
| <div class="line"><a name="l00441"></a><span class="lineno"> 441</span> CREATE FUNCTION MADLIB_SCHEMA.create_nb_classify_view(</div> |
| <div class="line"><a name="l00442"></a><span class="lineno"> 442</span>  "trainingSource" VARCHAR,</div> |
| <div class="line"><a name="l00443"></a><span class="lineno"> 443</span>  "trainingClassColumn" VARCHAR,</div> |
| <div class="line"><a name="l00444"></a><span class="lineno"> 444</span>  "trainingAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00445"></a><span class="lineno"> 445</span>  "classifySource" VARCHAR,</div> |
| <div class="line"><a name="l00446"></a><span class="lineno"> 446</span>  "classifyKeyColumn" VARCHAR,</div> |
| <div class="line"><a name="l00447"></a><span class="lineno"> 447</span>  "classifyAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00448"></a><span class="lineno"> 448</span>  "numAttrs" INTEGER,</div> |
| <div class="line"><a name="l00449"></a><span class="lineno"> 449</span>  "destName" VARCHAR)</div> |
| <div class="line"><a name="l00450"></a><span class="lineno"> 450</span> RETURNS VOID</div> |
| <div class="line"><a name="l00451"></a><span class="lineno"><a class="code" href="bayes_8sql__in.html#a798402280fc6db710957ae3ab58767e0"> 451</a></span> AS $$PythonFunction(bayes, bayes, create_classification_view)$$</div> |
| <div class="line"><a name="l00452"></a><span class="lineno"> 452</span> LANGUAGE plpythonu VOLATILE;</div> |
| <div class="line"><a name="l00453"></a><span class="lineno"> 453</span> </div> |
| <div class="line"><a name="l00454"></a><span class="lineno"> 454</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00455"></a><span class="lineno"> 455</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00456"></a><span class="lineno"> 456</span> <span class="comment"> * @brief Create view with columns <tt>(key, class, nb_prob)</tt></span></div> |
| <div class="line"><a name="l00457"></a><span class="lineno"> 457</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00458"></a><span class="lineno"> 458</span> <span class="comment"> * The created view will be of the following form:</span></div> |
| <div class="line"><a name="l00459"></a><span class="lineno"> 459</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00460"></a><span class="lineno"> 460</span> <span class="comment"> * <pre>VIEW <em>destName</em> (</span></div> |
| <div class="line"><a name="l00461"></a><span class="lineno"> 461</span> <span class="comment"> * key ANYTYPE,</span></div> |
| <div class="line"><a name="l00462"></a><span class="lineno"> 462</span> <span class="comment"> * class INTEGER,</span></div> |
| <div class="line"><a name="l00463"></a><span class="lineno"> 463</span> <span class="comment"> * nb_prob FLOAT8</span></div> |
| <div class="line"><a name="l00464"></a><span class="lineno"> 464</span> <span class="comment"> *)</pre></span></div> |
| <div class="line"><a name="l00465"></a><span class="lineno"> 465</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00466"></a><span class="lineno"> 466</span> <span class="comment"> * where \c nb_prob is the Naive-Bayes probability that \c class is the true</span></div> |
| <div class="line"><a name="l00467"></a><span class="lineno"> 467</span> <span class="comment"> * class of the record in \em classifySource identified by \c key.</span></div> |
| <div class="line"><a name="l00468"></a><span class="lineno"> 468</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00469"></a><span class="lineno"> 469</span> <span class="comment"> * @param featureProbsSource Name of table with precomputed feature</span></div> |
| <div class="line"><a name="l00470"></a><span class="lineno"> 470</span> <span class="comment"> * probabilities, as created with create_nb_prepared_data_tables()</span></div> |
| <div class="line"><a name="l00471"></a><span class="lineno"> 471</span> <span class="comment"> * @param classPriorsSource Name of table with precomputed class priors, as</span></div> |
| <div class="line"><a name="l00472"></a><span class="lineno"> 472</span> <span class="comment"> * created with create_nb_prepared_data_tables()</span></div> |
| <div class="line"><a name="l00473"></a><span class="lineno"> 473</span> <span class="comment"> * @param classifySource Name of the relation that contains data to be classified</span></div> |
| <div class="line"><a name="l00474"></a><span class="lineno"> 474</span> <span class="comment"> * @param classifyKeyColumn Name of column in \em classifySource that can</span></div> |
| <div class="line"><a name="l00475"></a><span class="lineno"> 475</span> <span class="comment"> * serve as unique identifier (the key of the source relation)</span></div> |
| <div class="line"><a name="l00476"></a><span class="lineno"> 476</span> <span class="comment"> * @param classifyAttrColumn Name of attributes-array column in \em classifySource</span></div> |
| <div class="line"><a name="l00477"></a><span class="lineno"> 477</span> <span class="comment"> * @param numAttrs Number of attributes to use for classification</span></div> |
| <div class="line"><a name="l00478"></a><span class="lineno"> 478</span> <span class="comment"> * @param destName Name of the view to create</span></div> |
| <div class="line"><a name="l00479"></a><span class="lineno"> 479</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00480"></a><span class="lineno"> 480</span> <span class="comment"> * @note \c create_nb_probs_view can be called in an ad-hoc fashion. See</span></div> |
| <div class="line"><a name="l00481"></a><span class="lineno"> 481</span> <span class="comment"> * \ref grp_bayes for instructions.</span></div> |
| <div class="line"><a name="l00482"></a><span class="lineno"> 482</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00483"></a><span class="lineno"> 483</span> <span class="comment"> * @usage</span></div> |
| <div class="line"><a name="l00484"></a><span class="lineno"> 484</span> <span class="comment"> * -# Create Naive Bayes probabilities view:</span></div> |
| <div class="line"><a name="l00485"></a><span class="lineno"> 485</span> <span class="comment"> * <pre>SELECT \ref create_nb_probs_view(</span></div> |
| <div class="line"><a name="l00486"></a><span class="lineno"> 486</span> <span class="comment"> * '<em>featureProbsName</em>', '<em>classPriorsName</em>',</span></div> |
| <div class="line"><a name="l00487"></a><span class="lineno"> 487</span> <span class="comment"> * '<em>classifySource</em>', '<em>classifyKeyColumn</em>', '<em>classifyAttrColumn</em>',</span></div> |
| <div class="line"><a name="l00488"></a><span class="lineno"> 488</span> <span class="comment"> * <em>numAttrs</em>, '<em>destName</em>'</span></div> |
| <div class="line"><a name="l00489"></a><span class="lineno"> 489</span> <span class="comment"> *);</pre></span></div> |
| <div class="line"><a name="l00490"></a><span class="lineno"> 490</span> <span class="comment"> * -# Show Naive Bayes probabilities:</span></div> |
| <div class="line"><a name="l00491"></a><span class="lineno"> 491</span> <span class="comment"> * <pre>SELECT * FROM <em>destName</em>;</pre></span></div> |
| <div class="line"><a name="l00492"></a><span class="lineno"> 492</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00493"></a><span class="lineno"> 493</span> <span class="comment"> * @internal</span></div> |
| <div class="line"><a name="l00494"></a><span class="lineno"> 494</span> <span class="comment"> * @sa This function is a wrapper for bayes::create_bayes_probabilities().</span></div> |
| <div class="line"><a name="l00495"></a><span class="lineno"> 495</span> <span class="comment"> */</span></div> |
| <div class="line"><a name="l00496"></a><span class="lineno"> 496</span> CREATE FUNCTION MADLIB_SCHEMA.create_nb_probs_view(</div> |
| <div class="line"><a name="l00497"></a><span class="lineno"> 497</span>  "featureProbsSource" VARCHAR,</div> |
| <div class="line"><a name="l00498"></a><span class="lineno"> 498</span>  "classPriorsSource" VARCHAR,</div> |
| <div class="line"><a name="l00499"></a><span class="lineno"> 499</span>  "classifySource" VARCHAR,</div> |
| <div class="line"><a name="l00500"></a><span class="lineno"> 500</span>  "classifyKeyColumn" VARCHAR,</div> |
| <div class="line"><a name="l00501"></a><span class="lineno"> 501</span>  "classifyAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00502"></a><span class="lineno"> 502</span>  "numAttrs" INTEGER,</div> |
| <div class="line"><a name="l00503"></a><span class="lineno"> 503</span>  "destName" VARCHAR)</div> |
| <div class="line"><a name="l00504"></a><span class="lineno"> 504</span> RETURNS VOID</div> |
| <div class="line"><a name="l00505"></a><span class="lineno"> 505</span> AS $$PythonFunction(bayes, bayes, create_bayes_probabilities_view)$$</div> |
| <div class="line"><a name="l00506"></a><span class="lineno"> 506</span> LANGUAGE plpythonu VOLATILE;</div> |
| <div class="line"><a name="l00507"></a><span class="lineno"> 507</span> </div> |
| <div class="line"><a name="l00508"></a><span class="lineno"> 508</span> CREATE FUNCTION MADLIB_SCHEMA.create_nb_probs_view(</div> |
| <div class="line"><a name="l00509"></a><span class="lineno"> 509</span>  "trainingSource" VARCHAR,</div> |
| <div class="line"><a name="l00510"></a><span class="lineno"> 510</span>  "trainingClassColumn" VARCHAR,</div> |
| <div class="line"><a name="l00511"></a><span class="lineno"> 511</span>  "trainingAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00512"></a><span class="lineno"> 512</span>  "classifySource" VARCHAR,</div> |
| <div class="line"><a name="l00513"></a><span class="lineno"> 513</span>  "classifyKeyColumn" VARCHAR,</div> |
| <div class="line"><a name="l00514"></a><span class="lineno"> 514</span>  "classifyAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00515"></a><span class="lineno"> 515</span>  "numAttrs" INTEGER,</div> |
| <div class="line"><a name="l00516"></a><span class="lineno"> 516</span>  "destName" VARCHAR)</div> |
| <div class="line"><a name="l00517"></a><span class="lineno"> 517</span> RETURNS VOID</div> |
| <div class="line"><a name="l00518"></a><span class="lineno"><a class="code" href="bayes_8sql__in.html#a163afffd0c845d325f060f74bcf02243"> 518</a></span> AS $$PythonFunction(bayes, bayes, create_bayes_probabilities_view)$$</div> |
| <div class="line"><a name="l00519"></a><span class="lineno"> 519</span> LANGUAGE plpythonu VOLATILE;</div> |
| <div class="line"><a name="l00520"></a><span class="lineno"> 520</span> </div> |
| <div class="line"><a name="l00521"></a><span class="lineno"> 521</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00522"></a><span class="lineno"> 522</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00523"></a><span class="lineno"> 523</span> <span class="comment"> * @brief Create a SQL function mapping arrays of attribute values to the Naive</span></div> |
| <div class="line"><a name="l00524"></a><span class="lineno"> 524</span> <span class="comment"> * Bayes classification.</span></div> |
| <div class="line"><a name="l00525"></a><span class="lineno"> 525</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00526"></a><span class="lineno"> 526</span> <span class="comment"> * The created SQL function is bound to the given feature probabilities and</span></div> |
| <div class="line"><a name="l00527"></a><span class="lineno"> 527</span> <span class="comment"> * class priors. Its declaration will be:</span></div> |
| <div class="line"><a name="l00528"></a><span class="lineno"> 528</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00529"></a><span class="lineno"> 529</span> <span class="comment"> * <tt></span></div> |
| <div class="line"><a name="l00530"></a><span class="lineno"> 530</span> <span class="comment"> * FUNCTION <em>destName</em> (attributes INTEGER[], smoothingFactor DOUBLE PRECISION)</span></div> |
| <div class="line"><a name="l00531"></a><span class="lineno"> 531</span> <span class="comment"> * RETURNS INTEGER[]</tt></span></div> |
| <div class="line"><a name="l00532"></a><span class="lineno"> 532</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00533"></a><span class="lineno"> 533</span> <span class="comment"> * The return type is \c INTEGER[] because the Naive Bayes classification might</span></div> |
| <div class="line"><a name="l00534"></a><span class="lineno"> 534</span> <span class="comment"> * be ambiguous (in which case all of the most likely candiates are returned).</span></div> |
| <div class="line"><a name="l00535"></a><span class="lineno"> 535</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00536"></a><span class="lineno"> 536</span> <span class="comment"> * @param featureProbsSource Name of table with precomputed feature</span></div> |
| <div class="line"><a name="l00537"></a><span class="lineno"> 537</span> <span class="comment"> * probabilities, as created with create_nb_prepared_data_tables()</span></div> |
| <div class="line"><a name="l00538"></a><span class="lineno"> 538</span> <span class="comment"> * @param classPriorsSource Name of table with precomputed class priors, as</span></div> |
| <div class="line"><a name="l00539"></a><span class="lineno"> 539</span> <span class="comment"> * created with create_nb_prepared_data_tables()</span></div> |
| <div class="line"><a name="l00540"></a><span class="lineno"> 540</span> <span class="comment"> * @param numAttrs Number of attributes to use for classification</span></div> |
| <div class="line"><a name="l00541"></a><span class="lineno"> 541</span> <span class="comment"> * @param destName Name of the function to create</span></div> |
| <div class="line"><a name="l00542"></a><span class="lineno"> 542</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00543"></a><span class="lineno"> 543</span> <span class="comment"> * @note</span></div> |
| <div class="line"><a name="l00544"></a><span class="lineno"> 544</span> <span class="comment"> * Just like \ref create_nb_classify_view and \ref create_nb_probs_view,</span></div> |
| <div class="line"><a name="l00545"></a><span class="lineno"> 545</span> <span class="comment"> * also \c create_nb_classify_fn can be called in an ad-hoc fashion. See</span></div> |
| <div class="line"><a name="l00546"></a><span class="lineno"> 546</span> <span class="comment"> * \ref grp_bayes for instructions.</span></div> |
| <div class="line"><a name="l00547"></a><span class="lineno"> 547</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00548"></a><span class="lineno"> 548</span> <span class="comment"> * @usage</span></div> |
| <div class="line"><a name="l00549"></a><span class="lineno"> 549</span> <span class="comment"> * -# Create classification function:</span></div> |
| <div class="line"><a name="l00550"></a><span class="lineno"> 550</span> <span class="comment"> * <pre>SELECT create_nb_classify_fn(</span></div> |
| <div class="line"><a name="l00551"></a><span class="lineno"> 551</span> <span class="comment"> * '<em>featureProbsSource</em>', '<em>classPriorsSource</em>',</span></div> |
| <div class="line"><a name="l00552"></a><span class="lineno"> 552</span> <span class="comment"> * <em>numAttrs</em>, '<em>destName</em>'</span></div> |
| <div class="line"><a name="l00553"></a><span class="lineno"> 553</span> <span class="comment"> *);</pre></span></div> |
| <div class="line"><a name="l00554"></a><span class="lineno"> 554</span> <span class="comment"> * -# Run classification function:</span></div> |
| <div class="line"><a name="l00555"></a><span class="lineno"> 555</span> <span class="comment"> * <pre>SELECT <em>destName</em>(<em>attributes</em>, <em>smoothingFactor</em>);</pre></span></div> |
| <div class="line"><a name="l00556"></a><span class="lineno"> 556</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00557"></a><span class="lineno"> 557</span> <span class="comment"> * @note</span></div> |
| <div class="line"><a name="l00558"></a><span class="lineno"> 558</span> <span class="comment"> * On Greenplum, the generated SQL function can only be called on the master.</span></div> |
| <div class="line"><a name="l00559"></a><span class="lineno"> 559</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00560"></a><span class="lineno"> 560</span> <span class="comment"> * @internal</span></div> |
| <div class="line"><a name="l00561"></a><span class="lineno"> 561</span> <span class="comment"> * @sa This function is a wrapper for bayes::create_classification_function().</span></div> |
| <div class="line"><a name="l00562"></a><span class="lineno"> 562</span> <span class="comment"> */</span></div> |
| <div class="line"><a name="l00563"></a><span class="lineno"> 563</span> CREATE FUNCTION MADLIB_SCHEMA.create_nb_classify_fn(</div> |
| <div class="line"><a name="l00564"></a><span class="lineno"> 564</span>  "featureProbsSource" VARCHAR,</div> |
| <div class="line"><a name="l00565"></a><span class="lineno"> 565</span>  "classPriorsSource" VARCHAR,</div> |
| <div class="line"><a name="l00566"></a><span class="lineno"> 566</span>  "numAttrs" INTEGER,</div> |
| <div class="line"><a name="l00567"></a><span class="lineno"> 567</span>  "destName" VARCHAR)</div> |
| <div class="line"><a name="l00568"></a><span class="lineno"> 568</span> RETURNS VOID</div> |
| <div class="line"><a name="l00569"></a><span class="lineno"> 569</span> AS $$PythonFunction(bayes, bayes, create_classification_function)$$</div> |
| <div class="line"><a name="l00570"></a><span class="lineno"> 570</span> LANGUAGE plpythonu VOLATILE;</div> |
| <div class="line"><a name="l00571"></a><span class="lineno"> 571</span> </div> |
| <div class="line"><a name="l00572"></a><span class="lineno"> 572</span> CREATE FUNCTION MADLIB_SCHEMA.create_nb_classify_fn(</div> |
| <div class="line"><a name="l00573"></a><span class="lineno"> 573</span>  "trainingSource" VARCHAR,</div> |
| <div class="line"><a name="l00574"></a><span class="lineno"> 574</span>  "trainingClassColumn" VARCHAR,</div> |
| <div class="line"><a name="l00575"></a><span class="lineno"> 575</span>  "trainingAttrColumn" VARCHAR,</div> |
| <div class="line"><a name="l00576"></a><span class="lineno"> 576</span>  "numAttrs" INTEGER,</div> |
| <div class="line"><a name="l00577"></a><span class="lineno"> 577</span>  "destName" VARCHAR)</div> |
| <div class="line"><a name="l00578"></a><span class="lineno"> 578</span> RETURNS VOID</div> |
| <div class="line"><a name="l00579"></a><span class="lineno"> 579</span> AS $$PythonFunction(bayes, bayes, create_classification_function)$$</div> |
| <div class="line"><a name="l00580"></a><span class="lineno"> 580</span> LANGUAGE plpythonu VOLATILE;</div> |
| </div><!-- fragment --></div><!-- contents --> |
| </div><!-- doc-content --> |
| <!-- start footer part --> |
| <div id="nav-path" class="navpath"><!-- id is needed for treeview function! --> |
| <ul> |
| <li class="navelem"><a class="el" href="dir_68267d1309a1af8e8297ef4c3efbcdba.html">src</a></li><li class="navelem"><a class="el" href="dir_efbcf68973d247bbf15f9eecae7f24e3.html">ports</a></li><li class="navelem"><a class="el" href="dir_a4a48839224ef8488facbffa8a397967.html">postgres</a></li><li class="navelem"><a class="el" href="dir_dc596537ad427a4d866006d1a3e1fe29.html">modules</a></li><li class="navelem"><a class="el" href="dir_3ee26ceeadb429f807bd6cd822eca1b9.html">bayes</a></li><li class="navelem"><a class="el" href="bayes_8sql__in.html">bayes.sql_in</a></li> |
| <li class="footer">Generated on Wed Aug 21 2013 16:09:52 for MADlib by |
| <a href="http://www.doxygen.org/index.html"> |
| <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.4 </li> |
| </ul> |
| </div> |
| </body> |
| </html> |
