public static void main(String[] arg) throws IOException {
// the file paths
String input = fileToPath("contextPasquier99.txt"); // the database
String output = ".//output.txt"; // the path for saving the frequent itemsets found
// minimum support
double minsup = 0.8; // means 2 transaction (we used a relative support)
// minimum bond
double minbond = 0.2; // the minimum bond threhsold
// Loading the transaction database
TransactionDatabase database = new TransactionDatabase();
try {
database.loadFile(input);
} catch (IOException e) {
e.printStackTrace();
}
// context.printContext();
// Applying the ECLAT algorithm
AlgoCORI algo = new AlgoCORI();
algo.runAlgorithm(output, database, minsup, minbond, false);
// if you change use "true" in the line above, ECLAT will use
// a triangular matrix for counting support of itemsets of size 2.
// For some datasets it should make the algorithm faster.
algo.printStats();
}
/**
* This method scans the database to calculate the support of each single item
*
* @param database the transaction database
* @param mapItemTIDS a map to store the tidset corresponding to each item
* @return the maximum item id appearing in this database
*/
int calculateSupportSingleItems(
TransactionDatabase database, final Map<Integer, BitSetSupport> mapItemTIDS) {
// (1) First database pass : calculate diffsets of each item.
int maxItemId = 0;
// for each transaction
for (int i = 0; i < database.size(); i++) {
// Add the transaction id to the set of all transaction ids
// for each item in that transaction
// For each item
for (Integer item : database.getTransactions().get(i)) {
// Get the current tidset of that item
BitSetSupport tids = mapItemTIDS.get(item);
// If none, then we create one
if (tids == null) {
tids = new BitSetSupport();
// For a new item, we sets all the bits of its diffset to true
tids.bitset.set(0, database.size(), true);
mapItemTIDS.put(item, tids);
// we remember the largest item seen until now
if (item > maxItemId) {
maxItemId = item;
}
}
// We set to false the bit corresponding to this transaction
// in the diffset of that item
tids.bitset.set(i, false);
// END DECLAT
// we increase the support of that item
tids.support++;
}
}
return maxItemId;
}
public static void main(String[] arg) throws IOException {
// Loading the binary context
String input = fileToPath("contextPasquier99.txt"); // the database
String output = ".//output.txt"; // the path for saving the frequent itemsets found
double minsup = 0.4; // means a minsup of 2 transaction (we used a relative support)
TransactionDatabase database = new TransactionDatabase();
try {
database.loadFile(input);
} catch (IOException e) {
e.printStackTrace();
}
// Applying the ECLAT algorithm
AlgoEclat_Bitset algo = new AlgoEclat_Bitset();
algo.runAlgorithm(output, database, minsup, false);
// NOTE 1: if you use "true" in the line above, CHARM will use
// a triangular matrix for counting support of itemsets of size 2.
// For some datasets it should make the algorithm faster.
algo.printStats();
}
