Naive Bayes(朴素贝叶斯算法)[分类算法]

Naïve Bayes(朴素贝叶斯)分类算法的实现

(1) 简介：

(2)   算法描述：

(3)

 <?php
 /*
 *Naive Bayes朴素贝叶斯算法（分类算法的实现）
 */

 /*
 *把.txt中的内容读到数组中保存
 *$filename:文件名称
 */
 //--------------------------------------------------------------------
 function  getFileContent($filename)
 {
     $array = array(null);
     $content = file_get_contents($filename);
     $result = explode("\r\n",$content);
     //print_r(count($result));
     for($j=0;$j<count($result);$j++)
     {
         //print_r($result[$j]."<br>");
         $con = explode(" ",$result[$j]);
         array_push($array,$con);
     }
     array_splice($array,0,1);
     return $array;
 }
 //--------------------------------------------------------------------

 /*
 *NaiveBayes朴素贝叶斯算法
 *$test:测试文本；$train：训练文本；$flagsyes：yes；$flagsno：no
 */
 //--------------------------------------------------------------------
 function  NaiveBayes($test,$train,$flagsyes,$flagsno)
 {
     $count_yes = 0;
     $num = count($train[0]);
     for($i=1;$i<count($train);$i++)
     {
         if($train[$i][$num-1]==$flagsyes)$count_yes++;
     }
     $p_yes = $count_yes / (count($train)-1);
     $p_no = 1- $p_yes;

     $count_no = count($train)-1 - $count_yes;

     for($i=1;$i<count($test)-1;$i++)
     {
         $testnumyes = 0;
         $testnumno = 0;
         for($j=1;$j<count($train);$j++)
         {
             if(($train[$j][$i]==$test[$i])&&($train[$j][count($test)-1]==$flagsyes))$testnumyes++;
             else if(($train[$j][$i]==$test[$i])&&($train[$j][count($test)-1]==$flagsno))$testnumno++;
         }

         $array_yes[$i] = $testnumyes / $count_yes ;
         $array_no[$i] = $testnumno / $count_no ;
 /*
         print_r($testnumyes."<br>");
         print_r($testnumno."<br>");
         print_r($count_yes."<br>");
         print_r($count_no."<br>");
         print_r($array_no[$i]."<br>");
 */
     }

     $py=1;
     $pn=1;
     for($i=1;$i<count($test)-1;$i++){
         $py *= $array_yes[$i];
         $pn *= $array_no[$i];
     }

     $py *= $p_yes;
     $pn *= $p_no;

     if($py>$pn)return $flagsyes;
     else return $flagsno;

 /*    print_r($py."<br>");
         print_r($pn."<br>");
 */    

 }
 //--------------------------------------------------------------------

 $train = getFileContent("train.txt");
 $test = getFileContent("test.txt");

 for($i=1;$i<count($test);$i++)
 {
     $test[$i][count($test[0])-1] = NaiveBayes($test[$i],$train,Y,N);
 }

 /*
 *将数组中的内容读到.txt中
 */
 //--------------------------------------------------------------------
 $fp= fopen('result.txt','wb');
 for($i=0;$i<count($test);$i++)
 {
     $temp = NULL;
     for($j=0;$j<count($test[$i]);$j++)
     {
         $temp =  $test[$i][$j]."\t";
         fwrite($fp,$temp);
     }
     fwrite($fp,"\r\n");
 }
 fclose($fp);
 //--------------------------------------------------------------------

 /*
 *打印输出
 */
 //--------------------------------------------------------------------
 echo "<pre>";
 print_r($test);
 echo "</pre>";
 //--------------------------------------------------------------------
 ?>