Aprori算法[关联规则算法]

<meta http-equiv="Content-Type" content="text/html; charset=GBK" />
<?php
/*
*Apriori算法（关联规则算法的实现）
*/

/*
*项目集X支持G个数小于等于1去除
*/
//--------------------------------------------------------------------
function first($train)
{
    $new_array = $train;
    $array_count = array(NULL);
    array_splice($array_count,0,1);
    for($i=1;$i<count($train[0]);$i++)
    {
        $count = 0;
        for($j=1;$j<count($train);$j++)
        {
            if($train[$j][$i]==1)
            {
                $count++;
            }
        }

        if($count==1){//非频繁项目集
            array_push($array_count,$i);
        }
    }

    for($i=count($array_count)-1;$i>=0;$i--)
    {
        for($j=0;$j<count($train);$j++)
        {
            array_splice($new_array[$j],$array_count[$i],1);
        }
            
    }
    
    return $new_array;
}
//--------------------------------------------------------------------

/*
*总事务D中包含事务X的数量
*/
//--------------------------------------------------------------------
function search($train,$array)
{
    $array_num = array();
    array_splice($array_num,0,1);
    for($i=0;$i<count($array);$i++)
    {
        for($j=1;$j<count($train[0]);$j++)
        {

            if($train[0][$j]==$array[$i])
            {
                array_push($array_num,$j);
                break;
            }
        }
    }
    $count = 0;//用于统计满足数据项集X的事务个数
    for($i=1;$i<count($train);$i++)
    {
        $flags = true;
        for($j=0;$j<count($array_num);$j++)
        {
            if($train[$i][$array_num[$j]]==0)
            {
                $flags = false;
                break;
            }
        }
        if($flags)$count++;
        
    }
    return $count;
}
//--------------------------------------------------------------------

/*
*合并两数组，除去其中相同的元素
*/
//--------------------------------------------------------------------
function  merge($arrayA,$arrayB)
{
    $array =  array(NULL);
    array_splice($array,0,1);
    
    //将两数组中的元素合并
    for($i=0;$i<count($arrayA)+count($arrayB);$i++)
    {
        if($i<count($arrayA))
        {
            array_push($array,$arrayA[$i]);
        }
        else {
            array_push($array,$arrayB[$i-count($arrayA)]);
        }
    }

    $array = array_unique($array);//删除两数组中重复的元素
    
    //对新的数组下标重新进行排序
    foreach($array as $value)
    {
        $newarray[]=$value;
    }
    return $newarray;
}
//--------------------------------------------------------------------

/*
*判断两个一维数组是否相等（不考虑键的位置，也不考虑数组中有重复的元素）
*/
//--------------------------------------------------------------------
function judgeequal($arrayA,$arrayB)
{

    if(count($arrayA)!=count($arrayB))
    {
        $flags = false;
    }else{
        $flags2 = true;
        for($i=0;$i<count($arrayA);$i++)
        {
            if(!in_array($arrayA[$i],$arrayB,true))
            {
                $flags2 = false;
                break;
            }
        }
        if($flags2)
        {
            $flags = true;
        }else {
            $flags = false;
        }
    }
    return $flags;
}
//--------------------------------------------------------------------

/*
*求支持度和置信度
*/
//--------------------------------------------------------------------
function support_confidence($arrayA,$arrayB,$train,&$support,&$confidence)
{
    $newarray = merge($arrayA,$arrayB);
    $countxy = search($train,$newarray);
    $support = $countxy / (count($train)-1);//项集X的支持度
    $countx = search($train,$arrayA);
    $confidence = $countxy / $countx;
    return 0;
    
}
//--------------------------------------------------------------------

/*
*Apriori算法
*/
//--------------------------------------------------------------------
function Apriori($train,$sup,$con)
{
    $aprioriAll = array();//存储所有的关联
    array_splice($aprioriAll,0,1);
    $apriori = array();
    array_splice($apriori,0,1);
    $train = first($train);
    for($i=1;$i<count($train[0])-1;$i++)
    {
        $arrayA[0] =$train[0][$i];
        for($j=$i+1;$j<count($train[0]);$j++)
        {
            $arrayB[0] =$train[0][$j];
            support_confidence($arrayA,$arrayB,$train,$support,$confidence);
            if(($support>=$sup)&&($confidence>=$con))
            {
                $Meg = merge($arrayA,$arrayB);
                array_push($apriori,$Meg);
            }
        }
    }
    array_push($aprioriAll,$apriori);
    while(count($apriori)>1)
    {
        $array = array();
        array_splice($array,0,1);
        for($i=0;$i<count($apriori)-1;$i++)
        {
            for($j=$i+1;$j<count($apriori);$j++)
            {
                $arrayB = merge($apriori[$i],$apriori[$j]);
                support_confidence($apriori[$i],$arrayB,$train,$support,$confidence);
                if(($support>=$sup)&&($confidence>=$con))
                {
/*                    echo "<pre>";
                    print_r($arrayA);
                    echo "<pre>";
                    print_r($arrayB);
*/                
                    array_push($array,$arrayB);
                }
                
            }
        }
        $apriori = $array;
        for($i=0;$i<count($apriori)-1;$i++)
        {
            for($j=count($apriori)-1;$j>=$i+1;$j--)
            {
                if(judgeequal($apriori[$i],$apriori[$j]))
                {
                    array_splice($apriori,$j,1);
                }
            }
        }
        foreach($apriori as $value)
        {
            $newarray[]=$value;
        }
        $apriori = $newarray;
        array_push($aprioriAll,$apriori);
    }
    return $aprioriAll;
}
//--------------------------------------------------------------------

/*
*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);
    $result = explode("
",$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;
}
//--------------------------------------------------------------------


/*
*把数组中内容写到.txt中保存
*$result:要存储的数组内容
*$filename:文件名称
*/
//--------------------------------------------------------------------
function Array_Totxt($result,$filename)
{
    $fp= fopen($filename,'wb');
    for($i=0;$i<count($result);$i++)
    {
        
        for($j=0;$j<count($result[$i]);$j++)
        {
            
            $temp = NULL;
            for($k=0;$k<count($result[$i][$j]);$k++){
            $temp =  $result[$i][$j][$k]."	";
            fwrite($fp,$temp);
            }
            fwrite($fp,"
");
        }
        
    }
    fclose($fp);
}
//--------------------------------------------------------------------
    $train = getFileContent("train.txt");
//    $train = getFileContent("er.txt");
    $apriori = Apriori($train,0.5,0.6);
    echo "<pre>";
    print_r($apriori);
    Array_Totxt($apriori,"result.txt")
//    Array_Totxt($apriori,"erresult.txt")
?>