main.cpp

apriori算法是数据挖掘的经典算法,它基于关联规则的思想.此为我的第3个收藏

/*=============================================================================
|
| Description: This program is an implementation of Apriori
|              learning algorithm.
|
| Usage:       webminer infofile
|
|              infofile  file containing informations about features and
|                        the test dataset (eg, weather.id3)
|
=============================================================================*/

#include <strings.h>
#include <stdio.h>
#include <fcntl.h>
#include <stdlib.h>
#include <math.h>
#include <string.h>

#include "apriori.h"
#include "rule.h"
#include "util.h"


int pagenum;

ItemSets *traindata = NULL;

Apriori *m_apriori;
AssociationRule *m_rules;


ItemSets *readin(char *infofile)
{
    FILE *fp;
    char line[4096];
    char *t, *s;
    int pageno, i;

    ItemSets *result;
    ItemSet *psession;

    if((fp = fopen(infofile, "rt")) == NULL)
    {
        printf(" Cannot open the data file %s !\n", infofile);
        exit(-9);
    }


    pagenum = -1;
    result = new ItemSets();
    
    while(fgets(line, 4096, fp) != NULL)
    {
        psession = new ItemSet();

        if (strchr(line, ',') != NULL)
        {
            s = (char *) new char[strlen(line) + 1];
            strcpy(s, line);  /* can't strtok arg1 directly in case it's const */
            if ((t = strtok(s, ",")) == NULL)
            {
                delete s;
                break;
            }

            for ( ; t != NULL; t = strtok((char *)NULL, ","))
            {

                pageno = atoi(t);

                if(pagenum < pageno)
                    pagenum = pageno;

                for(i = 0; i < psession->count; i++)
                    if(pageno < psession->itemof(i))
                        break;

                psession->insert(i, pageno);
            }

            delete s;
        }
        else
        {
            pageno = atoi(line);
            if(pagenum < pageno)
                pagenum = pageno;
            psession->add(pageno);
        }

        result->add(psession);

        delete psession;
    }

    fclose(fp);

    pagenum++;

    return(result);
}

main( int argc, char *argv[] )
{
    int traincnt = 0, i, j;
    double interestmax;
    char datafile[100], rules[100];

    /*
      Get command line arguments
    */
    if ( argc < 2 )
    {
        printf("Usage : apriori datafile \n\n ");
        exit(-9);
    }

    strcpy(datafile, argv[1]);
    for(i = strlen(datafile)-1; i >= 0; i--)
        if(datafile[i] == '.')
        {
            datafile[i] = '\0';
            break;
        }

    strcpy(rules, datafile);
    strcat(rules, ".rules");

    // Read in the structure of the web site from a map file
    printf("\nRead in data from  %s ... \n", argv[1]);
    traindata = readin(argv[1]);
    printf("  %d sessions have been read in !\n\n", traindata->count);


    printf("\nTo find all the large itemsets  ... \n");

    m_apriori = new Apriori();
    m_rules = new AssociationRule();

    m_apriori->setsupport(0.01);
    m_rules->setconfidence(0.5);

    // Finding large Itemsets from training data
    m_apriori->FindLargeItemSets(traindata);

    // Association Rules from extracted large item sets
    m_rules->m_LargeItemSets = m_apriori->m_Ls;


    printf("\nTo generate all the rules ... \n");
    m_rules->genrules();
    m_rules->writeto(rules);

    printf("  %d rules have been written to %s \n", m_rules->m_numRules, rules);


    // Clear off
    delete m_apriori;
    delete m_rules;

    delete traindata;

}