MSApriori.cpp

Go to the documentation of this file.

/***************************************************************************
                          MSApriori.cpp  -  description
                             -------------------
    begin                : cs dec 26 2002
    copyright            : (C) 2002 by Ferenc Bodon
    email                : bodon@mit.bme.hu
 ***************************************************************************/

#include "MSApriori.hpp"
#include <iostream>
#include <vector>
#include <set>


using namespace std;

void MSApriori::support( const itemtype& candidate_size )
{
   set<itemtype> basket;
   vector<itemtype> basket_v;
   if( store_input )
   {
      if (candidate_size == 2)
      {
         while( input_output_manager.read_in_a_line( basket ) )
         {
            input_output_manager.basket_recode( basket, basket_v );
            if (basket_v.size()>1) reduced_baskets[basket_v]++;
         }
      }
      for (map<vector<itemtype>,unsigned long>::iterator it = reduced_baskets.begin();
           it!=reduced_baskets.end();it++)
         msapriori_trie->find_candidate(it->first,candidate_size,it->second);
   }
   else while( input_output_manager.read_in_a_line( basket ) )
   {
      input_output_manager.basket_recode(basket, basket_v);
          msapriori_trie->find_candidate(basket_v,candidate_size);
   }
}

MSApriori::MSApriori( ifstream& basket_file, const char*  output_file_name, 
                  const bool store_input):
                  input_output_manager(basket_file, output_file_name ),
                  store_input(store_input)
{
}

void MSApriori::MSAPRIORI_alg( ifstream& mis_file, const double min_conf, 
                               const bool quiet, const unsigned long size_threshold )
{
   vector<double> mis_abs;
   if(!quiet) cout<<endl<<"\t\tFinding frequent itemsets..."<<endl<<endl;
   itemtype candidate_size=1;
   itemtype longest_path,longest_path_after_delete=1;
   if(!quiet)
      cout<<endl<<"Determining the support of the items and deleting infrequent ones!"<<endl;
   set< pair<itemtype, unsigned long> > support_of_items;
   unsigned long basket_number = input_output_manager.find_frequent_items( mis_file,
                                                                           support_of_items, mis_abs );
   msapriori_trie = new MSApriori_Trie( basket_number, mis_abs );
   msapriori_trie->insert_frequent_items( support_of_items );

//   apriori_trie->show_content();
//   getchar();
   longest_path_after_delete = msapriori_trie->longest_path();
//   apriori_trie->show_content();
//   getchar();
   longest_path = msapriori_trie->longest_path();
   candidate_size++;
   if(!quiet) cout<<endl<<"Genarating "<<candidate_size<<"-itemset candidates!"<<endl;
   msapriori_trie->candidate_generation( candidate_size-1 );
//   apriori_trie->show_content();
//   getchar();
   while( longest_path < msapriori_trie->longest_path() )
   {
      input_output_manager.rewind();
      if(!quiet) cout<<"Determining the support of the "<<candidate_size<<"-itemset candidates!"<<endl;
      support( candidate_size );
//      apriori_trie->show_content();
//      getchar();
      if(!quiet) cout<<"Deleting infrequent itemsets!"<<endl;
      msapriori_trie->delete_infrequent( candidate_size);
      longest_path_after_delete = msapriori_trie->longest_path();
//      apriori_trie->show_content();
//      getchar();
      if (candidate_size == size_threshold )
      {
         if(!quiet) cout<<"Size threshold is reached!"<<endl;
         break;
      }
      longest_path = msapriori_trie->longest_path();
      candidate_size++;
      if( !quiet ) cout<<endl<<"Genarating "<<candidate_size<<"-itemset candidates!"<<endl;
      msapriori_trie->candidate_generation(candidate_size-1);
//      apriori_trie->show_content_preorder();
//      getchar();
   }
   msapriori_trie->write_content_to_file( input_output_manager );
   if (min_conf)
   {
      if(!quiet) cout<<"\nGenerating association rules...!\n";
      msapriori_trie->association( min_conf, input_output_manager );
   }
   if(!quiet) cout<<"\nMining is done!\n";
}

MSApriori::~MSApriori()
{
   delete msapriori_trie;
}

---