<?xml version="1.0" encoding="ascii"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<title>Bio.LogisticRegression</title>
<link rel="stylesheet" href="epydoc.css" type="text/css" />
<script type="text/javascript" src="epydoc.js"></script>
</head>
<body bgcolor="white" text="black" link="blue" vlink="#204080"
alink="#204080">
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<th class="navbar" width="100%"></th>
</tr>
</table>
<table width="100%" cellpadding="0" cellspacing="0">
<tr valign="top">
<td width="100%">
<span class="breadcrumbs">
<a href="Bio-module.html">Package Bio</a> ::
Module LogisticRegression
</span>
</td>
<td>
<table cellpadding="0" cellspacing="0">
<!-- hide/show private -->
<tr><td align="right"><span class="options">[<a href="javascript:void(0);" class="privatelink"
onclick="toggle_private();">hide private</a>]</span></td></tr>
<tr><td align="right"><span class="options"
>[<a href="frames.html" target="_top">frames</a
>] | <a href="Bio.LogisticRegression-pysrc.html"
target="_top">no frames</a>]</span></td></tr>
</table>
</td>
</tr>
</table>
<h1 class="epydoc">Source Code for <a href="Bio.LogisticRegression-module.html">Module Bio.LogisticRegression</a></h1>
<pre class="py-src">
<a name="L1"></a><tt class="py-lineno"> 1</tt> <tt class="py-line"><tt class="py-comment">#!/usr/bin/env python</tt> </tt>
<a name="L2"></a><tt class="py-lineno"> 2</tt> <tt class="py-line"><tt class="py-comment"># This code is part of the Biopython distribution and governed by its</tt> </tt>
<a name="L3"></a><tt class="py-lineno"> 3</tt> <tt class="py-line"><tt class="py-comment"># license. Please see the LICENSE file that should have been included</tt> </tt>
<a name="L4"></a><tt class="py-lineno"> 4</tt> <tt class="py-line"><tt class="py-comment"># as part of this package.</tt> </tt>
<a name="L5"></a><tt class="py-lineno"> 5</tt> <tt class="py-line"><tt class="py-docstring">"""</tt> </tt>
<a name="L6"></a><tt class="py-lineno"> 6</tt> <tt class="py-line"><tt class="py-docstring">This module provides code for doing logistic regressions.</tt> </tt>
<a name="L7"></a><tt class="py-lineno"> 7</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L8"></a><tt class="py-lineno"> 8</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L9"></a><tt class="py-lineno"> 9</tt> <tt class="py-line"><tt class="py-docstring">Classes:</tt> </tt>
<a name="L10"></a><tt class="py-lineno"> 10</tt> <tt class="py-line"><tt class="py-docstring">LogisticRegression Holds information for a LogisticRegression classifier.</tt> </tt>
<a name="L11"></a><tt class="py-lineno"> 11</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L12"></a><tt class="py-lineno"> 12</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L13"></a><tt class="py-lineno"> 13</tt> <tt class="py-line"><tt class="py-docstring">Functions:</tt> </tt>
<a name="L14"></a><tt class="py-lineno"> 14</tt> <tt class="py-line"><tt class="py-docstring">train Train a new classifier.</tt> </tt>
<a name="L15"></a><tt class="py-lineno"> 15</tt> <tt class="py-line"><tt class="py-docstring">calculate Calculate the probabilities of each class, given an observation.</tt> </tt>
<a name="L16"></a><tt class="py-lineno"> 16</tt> <tt class="py-line"><tt class="py-docstring">classify Classify an observation into a class.</tt> </tt>
<a name="L17"></a><tt class="py-lineno"> 17</tt> <tt class="py-line"><tt class="py-docstring">"""</tt> </tt>
<a name="L18"></a><tt class="py-lineno"> 18</tt> <tt class="py-line"> </tt>
<a name="L19"></a><tt class="py-lineno"> 19</tt> <tt class="py-line"><tt class="py-keyword">import</tt> <tt class="py-name">numpy</tt> </tt>
<a name="L20"></a><tt class="py-lineno"> 20</tt> <tt class="py-line"><tt class="py-keyword">import</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">linalg</tt> </tt>
<a name="L21"></a><tt class="py-lineno"> 21</tt> <tt class="py-line"> </tt>
<a name="LogisticRegression"></a><div id="LogisticRegression-def"><a name="L22"></a><tt class="py-lineno"> 22</tt> <a class="py-toggle" href="#" id="LogisticRegression-toggle" onclick="return toggle('LogisticRegression');">-</a><tt class="py-line"><tt class="py-keyword">class</tt> <a class="py-def-name" href="Bio.LogisticRegression.LogisticRegression-class.html">LogisticRegression</a><tt class="py-op">(</tt><tt class="py-base-class">object</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="LogisticRegression-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="LogisticRegression-expanded"><a name="L23"></a><tt class="py-lineno"> 23</tt> <tt class="py-line"> <tt class="py-docstring">"""Holds information necessary to do logistic regression</tt> </tt>
<a name="L24"></a><tt class="py-lineno"> 24</tt> <tt class="py-line"><tt class="py-docstring"> classification.</tt> </tt>
<a name="L25"></a><tt class="py-lineno"> 25</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L26"></a><tt class="py-lineno"> 26</tt> <tt class="py-line"><tt class="py-docstring"> Members:</tt> </tt>
<a name="L27"></a><tt class="py-lineno"> 27</tt> <tt class="py-line"><tt class="py-docstring"> beta List of the weights for each dimension.</tt> </tt>
<a name="L28"></a><tt class="py-lineno"> 28</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L29"></a><tt class="py-lineno"> 29</tt> <tt class="py-line"><tt class="py-docstring"> """</tt> </tt>
<a name="LogisticRegression.__init__"></a><div id="LogisticRegression.__init__-def"><a name="L30"></a><tt class="py-lineno"> 30</tt> <a class="py-toggle" href="#" id="LogisticRegression.__init__-toggle" onclick="return toggle('LogisticRegression.__init__');">-</a><tt class="py-line"> <tt class="py-keyword">def</tt> <a class="py-def-name" href="Bio.LogisticRegression.LogisticRegression-class.html#__init__">__init__</a><tt class="py-op">(</tt><tt class="py-param">self</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="LogisticRegression.__init__-collapsed" style="display:none;" pad="+++" indent="++++++++"></div><div id="LogisticRegression.__init__-expanded"><a name="L31"></a><tt class="py-lineno"> 31</tt> <tt class="py-line"> <tt class="py-docstring">"""LogisticRegression()"""</tt> </tt>
<a name="L32"></a><tt class="py-lineno"> 32</tt> <tt class="py-line"> <tt class="py-name">self</tt><tt class="py-op">.</tt><tt class="py-name">beta</tt> <tt class="py-op">=</tt> <tt class="py-op">[</tt><tt class="py-op">]</tt> </tt>
</div></div><a name="L33"></a><tt class="py-lineno"> 33</tt> <tt class="py-line"> </tt>
<a name="train"></a><div id="train-def"><a name="L34"></a><tt class="py-lineno"> 34</tt> <a class="py-toggle" href="#" id="train-toggle" onclick="return toggle('train');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="Bio.LogisticRegression-module.html#train">train</a><tt class="py-op">(</tt><tt class="py-param">xs</tt><tt class="py-op">,</tt> <tt class="py-param">ys</tt><tt class="py-op">,</tt> <tt class="py-param">update_fn</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">,</tt> <tt class="py-param">typecode</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="train-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="train-expanded"><a name="L35"></a><tt class="py-lineno"> 35</tt> <tt class="py-line"> <tt class="py-docstring">"""train(xs, ys[, update_fn]) -> LogisticRegression</tt> </tt>
<a name="L36"></a><tt class="py-lineno"> 36</tt> <tt class="py-line"><tt class="py-docstring"> </tt> </tt>
<a name="L37"></a><tt class="py-lineno"> 37</tt> <tt class="py-line"><tt class="py-docstring"> Train a logistic regression classifier on a training set. xs is a</tt> </tt>
<a name="L38"></a><tt class="py-lineno"> 38</tt> <tt class="py-line"><tt class="py-docstring"> list of observations and ys is a list of the class assignments,</tt> </tt>
<a name="L39"></a><tt class="py-lineno"> 39</tt> <tt class="py-line"><tt class="py-docstring"> which should be 0 or 1. xs and ys should contain the same number</tt> </tt>
<a name="L40"></a><tt class="py-lineno"> 40</tt> <tt class="py-line"><tt class="py-docstring"> of elements. update_fn is an optional callback function that</tt> </tt>
<a name="L41"></a><tt class="py-lineno"> 41</tt> <tt class="py-line"><tt class="py-docstring"> takes as parameters that iteration number and log likelihood.</tt> </tt>
<a name="L42"></a><tt class="py-lineno"> 42</tt> <tt class="py-line"><tt class="py-docstring"> </tt> </tt>
<a name="L43"></a><tt class="py-lineno"> 43</tt> <tt class="py-line"><tt class="py-docstring"> """</tt> </tt>
<a name="L44"></a><tt class="py-lineno"> 44</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">len</tt><tt class="py-op">(</tt><tt class="py-name">xs</tt><tt class="py-op">)</tt> <tt class="py-op">!=</tt> <tt class="py-name">len</tt><tt class="py-op">(</tt><tt class="py-name">ys</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L45"></a><tt class="py-lineno"> 45</tt> <tt class="py-line"> <tt class="py-keyword">raise</tt> <tt class="py-name">ValueError</tt><tt class="py-op">(</tt><tt class="py-string">"xs and ys should be the same length."</tt><tt class="py-op">)</tt> </tt>
<a name="L46"></a><tt class="py-lineno"> 46</tt> <tt class="py-line"> <tt class="py-name">classes</tt> <tt class="py-op">=</tt> <tt id="link-0" class="py-name" targets="Method Bio.Nexus.Nexus.StepMatrix.set()=Bio.Nexus.Nexus.StepMatrix-class.html#set,Method Bio.SVDSuperimposer.SVDSuperimposer'.SVDSuperimposer.set()=Bio.SVDSuperimposer.SVDSuperimposer%27.SVDSuperimposer-class.html#set"><a title="Bio.Nexus.Nexus.StepMatrix.set
Bio.SVDSuperimposer.SVDSuperimposer'.SVDSuperimposer.set" class="py-name" href="#" onclick="return doclink('link-0', 'set', 'link-0');">set</a></tt><tt class="py-op">(</tt><tt class="py-name">ys</tt><tt class="py-op">)</tt> </tt>
<a name="L47"></a><tt class="py-lineno"> 47</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">classes</tt> <tt class="py-op">!=</tt> <tt id="link-1" class="py-name"><a title="Bio.Nexus.Nexus.StepMatrix.set
Bio.SVDSuperimposer.SVDSuperimposer'.SVDSuperimposer.set" class="py-name" href="#" onclick="return doclink('link-1', 'set', 'link-0');">set</a></tt><tt class="py-op">(</tt><tt class="py-op">[</tt><tt class="py-number">0</tt><tt class="py-op">,</tt> <tt class="py-number">1</tt><tt class="py-op">]</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L48"></a><tt class="py-lineno"> 48</tt> <tt class="py-line"> <tt class="py-keyword">raise</tt> <tt class="py-name">ValueError</tt><tt class="py-op">(</tt><tt class="py-string">"Classes should be 0's and 1's"</tt><tt class="py-op">)</tt> </tt>
<a name="L49"></a><tt class="py-lineno"> 49</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">typecode</tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L50"></a><tt class="py-lineno"> 50</tt> <tt class="py-line"> <tt class="py-name">typecode</tt> <tt class="py-op">=</tt> <tt class="py-string">'d'</tt> </tt>
<a name="L51"></a><tt class="py-lineno"> 51</tt> <tt class="py-line"> </tt>
<a name="L52"></a><tt class="py-lineno"> 52</tt> <tt class="py-line"> <tt class="py-comment"># Dimensionality of the data is the dimensionality of the</tt> </tt>
<a name="L53"></a><tt class="py-lineno"> 53</tt> <tt class="py-line"> <tt class="py-comment"># observations plus a constant dimension.</tt> </tt>
<a name="L54"></a><tt class="py-lineno"> 54</tt> <tt class="py-line"> <tt class="py-name">N</tt><tt class="py-op">,</tt> <tt class="py-name">ndims</tt> <tt class="py-op">=</tt> <tt class="py-name">len</tt><tt class="py-op">(</tt><tt class="py-name">xs</tt><tt class="py-op">)</tt><tt class="py-op">,</tt> <tt class="py-name">len</tt><tt class="py-op">(</tt><tt class="py-name">xs</tt><tt class="py-op">[</tt><tt class="py-number">0</tt><tt class="py-op">]</tt><tt class="py-op">)</tt> <tt class="py-op">+</tt> <tt class="py-number">1</tt> </tt>
<a name="L55"></a><tt class="py-lineno"> 55</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">N</tt><tt class="py-op">==</tt><tt class="py-number">0</tt> <tt class="py-keyword">or</tt> <tt class="py-name">ndims</tt><tt class="py-op">==</tt><tt class="py-number">1</tt><tt class="py-op">:</tt> </tt>
<a name="L56"></a><tt class="py-lineno"> 56</tt> <tt class="py-line"> <tt class="py-keyword">raise</tt> <tt class="py-name">ValueError</tt><tt class="py-op">(</tt><tt class="py-string">"No observations or observation of 0 dimension."</tt><tt class="py-op">)</tt> </tt>
<a name="L57"></a><tt class="py-lineno"> 57</tt> <tt class="py-line"> </tt>
<a name="L58"></a><tt class="py-lineno"> 58</tt> <tt class="py-line"> <tt class="py-comment"># Make an X array, with a constant first dimension.</tt> </tt>
<a name="L59"></a><tt class="py-lineno"> 59</tt> <tt class="py-line"> <tt class="py-name">X</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">ones</tt><tt class="py-op">(</tt><tt class="py-op">(</tt><tt class="py-name">N</tt><tt class="py-op">,</tt> <tt class="py-name">ndims</tt><tt class="py-op">)</tt><tt class="py-op">,</tt> <tt class="py-name">typecode</tt><tt class="py-op">)</tt> </tt>
<a name="L60"></a><tt class="py-lineno"> 60</tt> <tt class="py-line"> <tt class="py-name">X</tt><tt class="py-op">[</tt><tt class="py-op">:</tt><tt class="py-op">,</tt> <tt class="py-number">1</tt><tt class="py-op">:</tt><tt class="py-op">]</tt> <tt class="py-op">=</tt> <tt class="py-name">xs</tt> </tt>
<a name="L61"></a><tt class="py-lineno"> 61</tt> <tt class="py-line"> <tt class="py-name">Xt</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">transpose</tt><tt class="py-op">(</tt><tt class="py-name">X</tt><tt class="py-op">)</tt> </tt>
<a name="L62"></a><tt class="py-lineno"> 62</tt> <tt class="py-line"> <tt class="py-name">y</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">asarray</tt><tt class="py-op">(</tt><tt class="py-name">ys</tt><tt class="py-op">,</tt> <tt class="py-name">typecode</tt><tt class="py-op">)</tt> </tt>
<a name="L63"></a><tt class="py-lineno"> 63</tt> <tt class="py-line"> </tt>
<a name="L64"></a><tt class="py-lineno"> 64</tt> <tt class="py-line"> <tt class="py-comment"># Initialize the beta parameter to 0.</tt> </tt>
<a name="L65"></a><tt class="py-lineno"> 65</tt> <tt class="py-line"> <tt class="py-name">beta</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">zeros</tt><tt class="py-op">(</tt><tt class="py-name">ndims</tt><tt class="py-op">,</tt> <tt class="py-name">typecode</tt><tt class="py-op">)</tt> </tt>
<a name="L66"></a><tt class="py-lineno"> 66</tt> <tt class="py-line"> </tt>
<a name="L67"></a><tt class="py-lineno"> 67</tt> <tt class="py-line"> <tt id="link-2" class="py-name" targets="Variable Bio.MarkovModel.MAX_ITERATIONS=Bio.MarkovModel-module.html#MAX_ITERATIONS"><a title="Bio.MarkovModel.MAX_ITERATIONS" class="py-name" href="#" onclick="return doclink('link-2', 'MAX_ITERATIONS', 'link-2');">MAX_ITERATIONS</a></tt> <tt class="py-op">=</tt> <tt class="py-number">500</tt> </tt>
<a name="L68"></a><tt class="py-lineno"> 68</tt> <tt class="py-line"> <tt class="py-name">CONVERGE_THRESHOLD</tt> <tt class="py-op">=</tt> <tt class="py-number">0.01</tt> </tt>
<a name="L69"></a><tt class="py-lineno"> 69</tt> <tt class="py-line"> <tt class="py-name">stepsize</tt> <tt class="py-op">=</tt> <tt class="py-number">1.0</tt> </tt>
<a name="L70"></a><tt class="py-lineno"> 70</tt> <tt class="py-line"> <tt class="py-comment"># Now iterate using Newton-Raphson until the log-likelihoods</tt> </tt>
<a name="L71"></a><tt class="py-lineno"> 71</tt> <tt class="py-line"> <tt class="py-comment"># converge.</tt> </tt>
<a name="L72"></a><tt class="py-lineno"> 72</tt> <tt class="py-line"> <tt id="link-3" class="py-name" targets="Variable Bio.PDB.Polypeptide.i=Bio.PDB.Polypeptide-module.html#i"><a title="Bio.PDB.Polypeptide.i" class="py-name" href="#" onclick="return doclink('link-3', 'i', 'link-3');">i</a></tt> <tt class="py-op">=</tt> <tt class="py-number">0</tt> </tt>
<a name="L73"></a><tt class="py-lineno"> 73</tt> <tt class="py-line"> <tt class="py-name">old_beta</tt> <tt class="py-op">=</tt> <tt class="py-name">old_llik</tt> <tt class="py-op">=</tt> <tt class="py-name">None</tt> </tt>
<a name="L74"></a><tt class="py-lineno"> 74</tt> <tt class="py-line"> <tt class="py-keyword">while</tt> <tt id="link-4" class="py-name"><a title="Bio.PDB.Polypeptide.i" class="py-name" href="#" onclick="return doclink('link-4', 'i', 'link-3');">i</a></tt> <tt class="py-op"><</tt> <tt id="link-5" class="py-name"><a title="Bio.MarkovModel.MAX_ITERATIONS" class="py-name" href="#" onclick="return doclink('link-5', 'MAX_ITERATIONS', 'link-2');">MAX_ITERATIONS</a></tt><tt class="py-op">:</tt> </tt>
<a name="L75"></a><tt class="py-lineno"> 75</tt> <tt class="py-line"> <tt class="py-comment"># Calculate the probabilities. p = e^(beta X) / (1+e^(beta X))</tt> </tt>
<a name="L76"></a><tt class="py-lineno"> 76</tt> <tt class="py-line"> <tt class="py-name">ebetaX</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt id="link-6" class="py-name" targets="Variable Bio.Affy.CelFile.exp=Bio.Affy.CelFile-module.html#exp"><a title="Bio.Affy.CelFile.exp" class="py-name" href="#" onclick="return doclink('link-6', 'exp', 'link-6');">exp</a></tt><tt class="py-op">(</tt><tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">dot</tt><tt class="py-op">(</tt><tt class="py-name">beta</tt><tt class="py-op">,</tt> <tt class="py-name">Xt</tt><tt class="py-op">)</tt><tt class="py-op">)</tt> </tt>
<a name="L77"></a><tt class="py-lineno"> 77</tt> <tt class="py-line"> <tt class="py-name">p</tt> <tt class="py-op">=</tt> <tt class="py-name">ebetaX</tt> <tt class="py-op">/</tt> <tt class="py-op">(</tt><tt class="py-number">1</tt><tt class="py-op">+</tt><tt class="py-name">ebetaX</tt><tt class="py-op">)</tt> </tt>
<a name="L78"></a><tt class="py-lineno"> 78</tt> <tt class="py-line"> </tt>
<a name="L79"></a><tt class="py-lineno"> 79</tt> <tt class="py-line"> <tt class="py-comment"># Find the log likelihood score and see if I've converged.</tt> </tt>
<a name="L80"></a><tt class="py-lineno"> 80</tt> <tt class="py-line"> <tt class="py-name">logp</tt> <tt class="py-op">=</tt> <tt class="py-name">y</tt><tt class="py-op">*</tt><tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt id="link-7" class="py-name" targets="Variable Bio.Affy.CelFile.log=Bio.Affy.CelFile-module.html#log"><a title="Bio.Affy.CelFile.log" class="py-name" href="#" onclick="return doclink('link-7', 'log', 'link-7');">log</a></tt><tt class="py-op">(</tt><tt class="py-name">p</tt><tt class="py-op">)</tt> <tt class="py-op">+</tt> <tt class="py-op">(</tt><tt class="py-number">1</tt><tt class="py-op">-</tt><tt class="py-name">y</tt><tt class="py-op">)</tt><tt class="py-op">*</tt><tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt id="link-8" class="py-name"><a title="Bio.Affy.CelFile.log" class="py-name" href="#" onclick="return doclink('link-8', 'log', 'link-7');">log</a></tt><tt class="py-op">(</tt><tt class="py-number">1</tt><tt class="py-op">-</tt><tt class="py-name">p</tt><tt class="py-op">)</tt> </tt>
<a name="L81"></a><tt class="py-lineno"> 81</tt> <tt class="py-line"> <tt class="py-name">llik</tt> <tt class="py-op">=</tt> <tt id="link-9" class="py-name" targets="Method Bio.Nexus.Nexus.StepMatrix.sum()=Bio.Nexus.Nexus.StepMatrix-class.html#sum,Method Bio.SubsMat.SeqMat.sum()=Bio.SubsMat.SeqMat-class.html#sum"><a title="Bio.Nexus.Nexus.StepMatrix.sum
Bio.SubsMat.SeqMat.sum" class="py-name" href="#" onclick="return doclink('link-9', 'sum', 'link-9');">sum</a></tt><tt class="py-op">(</tt><tt class="py-name">logp</tt><tt class="py-op">)</tt> </tt>
<a name="L82"></a><tt class="py-lineno"> 82</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">update_fn</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L83"></a><tt class="py-lineno"> 83</tt> <tt class="py-line"> <tt class="py-name">update_fn</tt><tt class="py-op">(</tt><tt class="py-name">iter</tt><tt class="py-op">,</tt> <tt class="py-name">llik</tt><tt class="py-op">)</tt> </tt>
<a name="L84"></a><tt class="py-lineno"> 84</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">old_llik</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L85"></a><tt class="py-lineno"> 85</tt> <tt class="py-line"> <tt class="py-comment"># Check to see if the likelihood decreased. If it did, then</tt> </tt>
<a name="L86"></a><tt class="py-lineno"> 86</tt> <tt class="py-line"> <tt class="py-comment"># restore the old beta parameters and half the step size.</tt> </tt>
<a name="L87"></a><tt class="py-lineno"> 87</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">llik</tt> <tt class="py-op"><</tt> <tt class="py-name">old_llik</tt><tt class="py-op">:</tt> </tt>
<a name="L88"></a><tt class="py-lineno"> 88</tt> <tt class="py-line"> <tt class="py-name">stepsize</tt> <tt class="py-op">=</tt> <tt class="py-name">stepsize</tt> <tt class="py-op">/</tt> <tt class="py-number">2.0</tt> </tt>
<a name="L89"></a><tt class="py-lineno"> 89</tt> <tt class="py-line"> <tt class="py-name">beta</tt> <tt class="py-op">=</tt> <tt class="py-name">old_beta</tt> </tt>
<a name="L90"></a><tt class="py-lineno"> 90</tt> <tt class="py-line"> <tt class="py-comment"># If I've converged, then stop.</tt> </tt>
<a name="L91"></a><tt class="py-lineno"> 91</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt id="link-10" class="py-name" targets="Variable Bio.Affy.CelFile.fabs=Bio.Affy.CelFile-module.html#fabs"><a title="Bio.Affy.CelFile.fabs" class="py-name" href="#" onclick="return doclink('link-10', 'fabs', 'link-10');">fabs</a></tt><tt class="py-op">(</tt><tt class="py-name">llik</tt><tt class="py-op">-</tt><tt class="py-name">old_llik</tt><tt class="py-op">)</tt> <tt class="py-op"><=</tt> <tt class="py-name">CONVERGE_THRESHOLD</tt><tt class="py-op">:</tt> </tt>
<a name="L92"></a><tt class="py-lineno"> 92</tt> <tt class="py-line"> <tt class="py-keyword">break</tt> </tt>
<a name="L93"></a><tt class="py-lineno"> 93</tt> <tt class="py-line"> <tt class="py-name">old_llik</tt><tt class="py-op">,</tt> <tt class="py-name">old_beta</tt> <tt class="py-op">=</tt> <tt class="py-name">llik</tt><tt class="py-op">,</tt> <tt class="py-name">beta</tt> </tt>
<a name="L94"></a><tt class="py-lineno"> 94</tt> <tt class="py-line"> <tt id="link-11" class="py-name"><a title="Bio.PDB.Polypeptide.i" class="py-name" href="#" onclick="return doclink('link-11', 'i', 'link-3');">i</a></tt> <tt class="py-op">+=</tt> <tt class="py-number">1</tt> </tt>
<a name="L95"></a><tt class="py-lineno"> 95</tt> <tt class="py-line"> </tt>
<a name="L96"></a><tt class="py-lineno"> 96</tt> <tt class="py-line"> <tt class="py-name">W</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">identity</tt><tt class="py-op">(</tt><tt class="py-name">N</tt><tt class="py-op">)</tt> <tt class="py-op">*</tt> <tt class="py-name">p</tt> </tt>
<a name="L97"></a><tt class="py-lineno"> 97</tt> <tt class="py-line"> <tt class="py-name">Xtyp</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">dot</tt><tt class="py-op">(</tt><tt class="py-name">Xt</tt><tt class="py-op">,</tt> <tt class="py-name">y</tt><tt class="py-op">-</tt><tt class="py-name">p</tt><tt class="py-op">)</tt> <tt class="py-comment"># Calculate the first derivative.</tt> </tt>
<a name="L98"></a><tt class="py-lineno"> 98</tt> <tt class="py-line"> <tt class="py-name">XtWX</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">dot</tt><tt class="py-op">(</tt><tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">dot</tt><tt class="py-op">(</tt><tt class="py-name">Xt</tt><tt class="py-op">,</tt> <tt class="py-name">W</tt><tt class="py-op">)</tt><tt class="py-op">,</tt> <tt class="py-name">X</tt><tt class="py-op">)</tt> <tt class="py-comment"># Calculate the second derivative.</tt> </tt>
<a name="L99"></a><tt class="py-lineno"> 99</tt> <tt class="py-line"> <tt class="py-comment">#u, s, vt = singular_value_decomposition(XtWX)</tt> </tt>
<a name="L100"></a><tt class="py-lineno">100</tt> <tt class="py-line"> <tt class="py-comment">#print "U", u</tt> </tt>
<a name="L101"></a><tt class="py-lineno">101</tt> <tt class="py-line"> <tt class="py-comment">#print "S", s</tt> </tt>
<a name="L102"></a><tt class="py-lineno">102</tt> <tt class="py-line"> <tt class="py-name">delta</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">linalg</tt><tt class="py-op">.</tt><tt class="py-name">solve</tt><tt class="py-op">(</tt><tt class="py-name">XtWX</tt><tt class="py-op">,</tt> <tt class="py-name">Xtyp</tt><tt class="py-op">)</tt> </tt>
<a name="L103"></a><tt class="py-lineno">103</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt id="link-12" class="py-name"><a title="Bio.Affy.CelFile.fabs" class="py-name" href="#" onclick="return doclink('link-12', 'fabs', 'link-10');">fabs</a></tt><tt class="py-op">(</tt><tt class="py-name">stepsize</tt><tt class="py-op">-</tt><tt class="py-number">1.0</tt><tt class="py-op">)</tt> <tt class="py-op">></tt> <tt class="py-number">0.001</tt><tt class="py-op">:</tt> </tt>
<a name="L104"></a><tt class="py-lineno">104</tt> <tt class="py-line"> <tt class="py-name">delta</tt> <tt class="py-op">=</tt> <tt class="py-name">delta</tt> <tt class="py-op">*</tt> <tt class="py-name">stepsize</tt> </tt>
<a name="L105"></a><tt class="py-lineno">105</tt> <tt class="py-line"> <tt class="py-name">beta</tt> <tt class="py-op">=</tt> <tt class="py-name">beta</tt> <tt class="py-op">+</tt> <tt class="py-name">delta</tt> <tt class="py-comment"># Update beta.</tt> </tt>
<a name="L106"></a><tt class="py-lineno">106</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L107"></a><tt class="py-lineno">107</tt> <tt class="py-line"> <tt class="py-keyword">raise</tt> <tt class="py-name">RuntimeError</tt><tt class="py-op">(</tt><tt class="py-string">"Didn't converge."</tt><tt class="py-op">)</tt> </tt>
<a name="L108"></a><tt class="py-lineno">108</tt> <tt class="py-line"> </tt>
<a name="L109"></a><tt class="py-lineno">109</tt> <tt class="py-line"> <tt class="py-name">lr</tt> <tt class="py-op">=</tt> <tt id="link-13" class="py-name" targets="Module Bio.LogisticRegression=Bio.LogisticRegression-module.html,Class Bio.LogisticRegression.LogisticRegression=Bio.LogisticRegression.LogisticRegression-class.html"><a title="Bio.LogisticRegression
Bio.LogisticRegression.LogisticRegression" class="py-name" href="#" onclick="return doclink('link-13', 'LogisticRegression', 'link-13');">LogisticRegression</a></tt><tt class="py-op">(</tt><tt class="py-op">)</tt> </tt>
<a name="L110"></a><tt class="py-lineno">110</tt> <tt class="py-line"> <tt class="py-name">lr</tt><tt class="py-op">.</tt><tt class="py-name">beta</tt> <tt class="py-op">=</tt> <tt class="py-name">map</tt><tt class="py-op">(</tt><tt class="py-name">float</tt><tt class="py-op">,</tt> <tt class="py-name">beta</tt><tt class="py-op">)</tt> <tt class="py-comment"># Convert back to regular array.</tt> </tt>
<a name="L111"></a><tt class="py-lineno">111</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">lr</tt> </tt>
</div><a name="L112"></a><tt class="py-lineno">112</tt> <tt class="py-line"> </tt>
<a name="calculate"></a><div id="calculate-def"><a name="L113"></a><tt class="py-lineno">113</tt> <a class="py-toggle" href="#" id="calculate-toggle" onclick="return toggle('calculate');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="Bio.LogisticRegression-module.html#calculate">calculate</a><tt class="py-op">(</tt><tt class="py-param">lr</tt><tt class="py-op">,</tt> <tt class="py-param">x</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="calculate-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="calculate-expanded"><a name="L114"></a><tt class="py-lineno">114</tt> <tt class="py-line"> <tt class="py-docstring">"""calculate(lr, x) -> list of probabilities</tt> </tt>
<a name="L115"></a><tt class="py-lineno">115</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L116"></a><tt class="py-lineno">116</tt> <tt class="py-line"><tt class="py-docstring"> Calculate the probability for each class. lr is a</tt> </tt>
<a name="L117"></a><tt class="py-lineno">117</tt> <tt class="py-line"><tt class="py-docstring"> LogisticRegression object. x is the observed data. Returns a</tt> </tt>
<a name="L118"></a><tt class="py-lineno">118</tt> <tt class="py-line"><tt class="py-docstring"> list of the probability that it fits each class.</tt> </tt>
<a name="L119"></a><tt class="py-lineno">119</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L120"></a><tt class="py-lineno">120</tt> <tt class="py-line"><tt class="py-docstring"> """</tt> </tt>
<a name="L121"></a><tt class="py-lineno">121</tt> <tt class="py-line"> <tt class="py-comment"># Insert a constant term for x.</tt> </tt>
<a name="L122"></a><tt class="py-lineno">122</tt> <tt class="py-line"> <tt id="link-14" class="py-name" targets="Variable Bio.Statistics.lowess.x=Bio.Statistics.lowess-module.html#x"><a title="Bio.Statistics.lowess.x" class="py-name" href="#" onclick="return doclink('link-14', 'x', 'link-14');">x</a></tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">asarray</tt><tt class="py-op">(</tt><tt class="py-op">[</tt><tt class="py-number">1.0</tt><tt class="py-op">]</tt> <tt class="py-op">+</tt> <tt id="link-15" class="py-name"><a title="Bio.Statistics.lowess.x" class="py-name" href="#" onclick="return doclink('link-15', 'x', 'link-14');">x</a></tt><tt class="py-op">)</tt> </tt>
<a name="L123"></a><tt class="py-lineno">123</tt> <tt class="py-line"> <tt class="py-comment"># Calculate the probability. p = e^(beta X) / (1+e^(beta X))</tt> </tt>
<a name="L124"></a><tt class="py-lineno">124</tt> <tt class="py-line"> <tt class="py-name">ebetaX</tt> <tt class="py-op">=</tt> <tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt id="link-16" class="py-name"><a title="Bio.Affy.CelFile.exp" class="py-name" href="#" onclick="return doclink('link-16', 'exp', 'link-6');">exp</a></tt><tt class="py-op">(</tt><tt class="py-name">numpy</tt><tt class="py-op">.</tt><tt class="py-name">dot</tt><tt class="py-op">(</tt><tt class="py-name">lr</tt><tt class="py-op">.</tt><tt class="py-name">beta</tt><tt class="py-op">,</tt> <tt id="link-17" class="py-name"><a title="Bio.Statistics.lowess.x" class="py-name" href="#" onclick="return doclink('link-17', 'x', 'link-14');">x</a></tt><tt class="py-op">)</tt><tt class="py-op">)</tt> </tt>
<a name="L125"></a><tt class="py-lineno">125</tt> <tt class="py-line"> <tt class="py-name">p</tt> <tt class="py-op">=</tt> <tt class="py-name">ebetaX</tt> <tt class="py-op">/</tt> <tt class="py-op">(</tt><tt class="py-number">1</tt><tt class="py-op">+</tt><tt class="py-name">ebetaX</tt><tt class="py-op">)</tt> </tt>
<a name="L126"></a><tt class="py-lineno">126</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-op">[</tt><tt class="py-number">1</tt><tt class="py-op">-</tt><tt class="py-name">p</tt><tt class="py-op">,</tt> <tt class="py-name">p</tt><tt class="py-op">]</tt> </tt>
</div><a name="L127"></a><tt class="py-lineno">127</tt> <tt class="py-line"> </tt>
<a name="classify"></a><div id="classify-def"><a name="L128"></a><tt class="py-lineno">128</tt> <a class="py-toggle" href="#" id="classify-toggle" onclick="return toggle('classify');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="Bio.LogisticRegression-module.html#classify">classify</a><tt class="py-op">(</tt><tt class="py-param">lr</tt><tt class="py-op">,</tt> <tt class="py-param">x</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="classify-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="classify-expanded"><a name="L129"></a><tt class="py-lineno">129</tt> <tt class="py-line"> <tt class="py-docstring">"""classify(lr, x) -> 1 or 0</tt> </tt>
<a name="L130"></a><tt class="py-lineno">130</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L131"></a><tt class="py-lineno">131</tt> <tt class="py-line"><tt class="py-docstring"> Classify an observation into a class.</tt> </tt>
<a name="L132"></a><tt class="py-lineno">132</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L133"></a><tt class="py-lineno">133</tt> <tt class="py-line"><tt class="py-docstring"> """</tt> </tt>
<a name="L134"></a><tt class="py-lineno">134</tt> <tt class="py-line"> <tt class="py-name">probs</tt> <tt class="py-op">=</tt> <tt id="link-18" class="py-name" targets="Function Bio.LogisticRegression.calculate()=Bio.LogisticRegression-module.html#calculate,Function Bio.MaxEntropy.calculate()=Bio.MaxEntropy-module.html#calculate,Function Bio.NaiveBayes.calculate()=Bio.NaiveBayes-module.html#calculate,Function Bio.kNN.calculate()=Bio.kNN-module.html#calculate"><a title="Bio.LogisticRegression.calculate
Bio.MaxEntropy.calculate
Bio.NaiveBayes.calculate
Bio.kNN.calculate" class="py-name" href="#" onclick="return doclink('link-18', 'calculate', 'link-18');">calculate</a></tt><tt class="py-op">(</tt><tt class="py-name">lr</tt><tt class="py-op">,</tt> <tt id="link-19" class="py-name"><a title="Bio.Statistics.lowess.x" class="py-name" href="#" onclick="return doclink('link-19', 'x', 'link-14');">x</a></tt><tt class="py-op">)</tt> </tt>
<a name="L135"></a><tt class="py-lineno">135</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">probs</tt><tt class="py-op">[</tt><tt class="py-number">0</tt><tt class="py-op">]</tt> <tt class="py-op">></tt> <tt class="py-name">probs</tt><tt class="py-op">[</tt><tt class="py-number">1</tt><tt class="py-op">]</tt><tt class="py-op">:</tt> </tt>
<a name="L136"></a><tt class="py-lineno">136</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-number">0</tt> </tt>
<a name="L137"></a><tt class="py-lineno">137</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-number">1</tt> </tt>
</div><a name="L138"></a><tt class="py-lineno">138</tt> <tt class="py-line"> </tt><script type="text/javascript">
<!--
expandto(location.href);
// -->
</script>
</pre>
<br />
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<th class="navbar" width="100%"></th>
</tr>
</table>
<table border="0" cellpadding="0" cellspacing="0" width="100%%">
<tr>
<td align="left" class="footer">
Generated by Epydoc 3.0.1 on Thu Aug 18 18:22:57 2011
</td>
<td align="right" class="footer">
<a target="mainFrame" href="http://epydoc.sourceforge.net"
>http://epydoc.sourceforge.net</a>
</td>
</tr>
</table>
<script type="text/javascript">
<!--
// Private objects are initially displayed (because if
// javascript is turned off then we want them to be
// visible); but by default, we want to hide them. So hide
// them unless we have a cookie that says to show them.
checkCookie();
// -->
</script>
</body>
</html>
