/**
* Subtracts an item set from another one.
*
* @param toSubtract the item set to be subtracted from this one.
* @return an item set that only contains items form this item sets that are not contained by
* toSubtract
*/
public final AprioriItemSet subtract(AprioriItemSet toSubtract) {
AprioriItemSet result = new AprioriItemSet(m_totalTransactions);
result.m_items = new int[m_items.length];
for (int i = 0; i < m_items.length; i++)
if (toSubtract.m_items[i] == -1) result.m_items[i] = m_items[i];
else result.m_items[i] = -1;
result.m_counter = 0;
return result;
}
/**
* Generates rules with more than one item in the consequence.
*
* @param rules all the rules having (k-1)-item sets as consequences
* @param numItemsInSet the size of the item set for which the rules are to be generated
* @param numItemsInConsequence the value of (k-1)
* @param minConfidence the minimum confidence a rule has to have
* @param hashtables the hashtables containing all(!) previously generated item sets
* @return all the rules having (k)-item sets as consequences
*/
private final FastVector[] moreComplexRules(
FastVector[] rules,
int numItemsInSet,
int numItemsInConsequence,
double minConfidence,
FastVector hashtables) {
AprioriItemSet newPremise;
FastVector[] result, moreResults;
FastVector newConsequences, newPremises = new FastVector(), newConf = new FastVector();
Hashtable hashtable;
if (numItemsInSet > numItemsInConsequence + 1) {
hashtable = (Hashtable) hashtables.elementAt(numItemsInSet - numItemsInConsequence - 2);
newConsequences = mergeAllItemSets(rules[1], numItemsInConsequence - 1, m_totalTransactions);
Enumeration enu = newConsequences.elements();
while (enu.hasMoreElements()) {
AprioriItemSet current = (AprioriItemSet) enu.nextElement();
current.m_counter = m_counter;
newPremise = subtract(current);
newPremise.m_counter = ((Integer) hashtable.get(newPremise)).intValue();
newPremises.addElement(newPremise);
newConf.addElement(new Double(confidenceForRule(newPremise, current)));
}
result = new FastVector[3];
result[0] = newPremises;
result[1] = newConsequences;
result[2] = newConf;
pruneRules(result, minConfidence);
moreResults =
moreComplexRules(
result, numItemsInSet, numItemsInConsequence + 1, minConfidence, hashtables);
if (moreResults != null)
for (int i = 0; i < moreResults[0].size(); i++) {
result[0].addElement(moreResults[0].elementAt(i));
result[1].addElement(moreResults[1].elementAt(i));
result[2].addElement(moreResults[2].elementAt(i));
}
return result;
} else return null;
}
/**
* Generates all rules for an item set.
*
* @param minConfidence the minimum confidence the rules have to have
* @param hashtables containing all(!) previously generated item sets
* @param numItemsInSet the size of the item set for which the rules are to be generated
* @return all the rules with minimum confidence for the given item set
*/
public FastVector[] generateRules(
double minConfidence, FastVector hashtables, int numItemsInSet) {
FastVector premises = new FastVector(),
consequences = new FastVector(),
conf = new FastVector();
FastVector[] rules = new FastVector[3], moreResults;
AprioriItemSet premise, consequence;
Hashtable hashtable = (Hashtable) hashtables.elementAt(numItemsInSet - 2);
// Generate all rules with one item in the consequence.
for (int i = 0; i < m_items.length; i++)
if (m_items[i] != -1) {
premise = new AprioriItemSet(m_totalTransactions);
consequence = new AprioriItemSet(m_totalTransactions);
premise.m_items = new int[m_items.length];
consequence.m_items = new int[m_items.length];
consequence.m_counter = m_counter;
for (int j = 0; j < m_items.length; j++) consequence.m_items[j] = -1;
System.arraycopy(m_items, 0, premise.m_items, 0, m_items.length);
premise.m_items[i] = -1;
consequence.m_items[i] = m_items[i];
premise.m_counter = ((Integer) hashtable.get(premise)).intValue();
premises.addElement(premise);
consequences.addElement(consequence);
conf.addElement(new Double(confidenceForRule(premise, consequence)));
}
rules[0] = premises;
rules[1] = consequences;
rules[2] = conf;
pruneRules(rules, minConfidence);
// Generate all the other rules
moreResults = moreComplexRules(rules, numItemsInSet, 1, minConfidence, hashtables);
if (moreResults != null)
for (int i = 0; i < moreResults[0].size(); i++) {
rules[0].addElement(moreResults[0].elementAt(i));
rules[1].addElement(moreResults[1].elementAt(i));
rules[2].addElement(moreResults[2].elementAt(i));
}
return rules;
}
/**
* Generates all significant rules for an item set.
*
* @param minMetric the minimum metric (confidence, lift, leverage, improvement) the rules have to
* have
* @param metricType (confidence=0, lift, leverage, improvement)
* @param hashtables containing all(!) previously generated item sets
* @param numItemsInSet the size of the item set for which the rules are to be generated
* @param numTransactions
* @param significanceLevel the significance level for testing the rules
* @return all the rules with minimum metric for the given item set
* @exception Exception if something goes wrong
*/
public final FastVector[] generateRulesBruteForce(
double minMetric,
int metricType,
FastVector hashtables,
int numItemsInSet,
int numTransactions,
double significanceLevel)
throws Exception {
FastVector premises = new FastVector(),
consequences = new FastVector(),
conf = new FastVector(),
lift = new FastVector(),
lev = new FastVector(),
conv = new FastVector();
FastVector[] rules = new FastVector[6];
AprioriItemSet premise, consequence;
Hashtable hashtableForPremise, hashtableForConsequence;
int numItemsInPremise, help, max, consequenceUnconditionedCounter;
double[][] contingencyTable = new double[2][2];
double metric, chiSquared = 0;
// Generate all possible rules for this item set and test their
// significance.
max = (int) Math.pow(2, numItemsInSet);
for (int j = 1; j < max; j++) {
numItemsInPremise = 0;
help = j;
while (help > 0) {
if (help % 2 == 1) numItemsInPremise++;
help /= 2;
}
if (numItemsInPremise < numItemsInSet) {
hashtableForPremise = (Hashtable) hashtables.elementAt(numItemsInPremise - 1);
hashtableForConsequence =
(Hashtable) hashtables.elementAt(numItemsInSet - numItemsInPremise - 1);
premise = new AprioriItemSet(m_totalTransactions);
consequence = new AprioriItemSet(m_totalTransactions);
premise.m_items = new int[m_items.length];
consequence.m_items = new int[m_items.length];
consequence.m_counter = m_counter;
help = j;
for (int i = 0; i < m_items.length; i++)
if (m_items[i] != -1) {
if (help % 2 == 1) {
premise.m_items[i] = m_items[i];
consequence.m_items[i] = -1;
} else {
premise.m_items[i] = -1;
consequence.m_items[i] = m_items[i];
}
help /= 2;
} else {
premise.m_items[i] = -1;
consequence.m_items[i] = -1;
}
premise.m_counter = ((Integer) hashtableForPremise.get(premise)).intValue();
consequenceUnconditionedCounter =
((Integer) hashtableForConsequence.get(consequence)).intValue();
if (significanceLevel != -1) {
contingencyTable[0][0] = (consequence.m_counter);
contingencyTable[0][1] = (premise.m_counter - consequence.m_counter);
contingencyTable[1][0] = (consequenceUnconditionedCounter - consequence.m_counter);
contingencyTable[1][1] =
(numTransactions
- premise.m_counter
- consequenceUnconditionedCounter
+ consequence.m_counter);
chiSquared = ContingencyTables.chiSquared(contingencyTable, false);
}
if (metricType == 0) {
metric = confidenceForRule(premise, consequence);
if ((!(metric < minMetric))
&& (significanceLevel == -1 || !(chiSquared > significanceLevel))) {
premises.addElement(premise);
consequences.addElement(consequence);
conf.addElement(new Double(metric));
lift.addElement(
new Double(liftForRule(premise, consequence, consequenceUnconditionedCounter)));
lev.addElement(
new Double(
leverageForRule(
premise, consequence, premise.m_counter, consequenceUnconditionedCounter)));
conv.addElement(
new Double(
convictionForRule(
premise, consequence, premise.m_counter, consequenceUnconditionedCounter)));
}
} else {
double tempConf = confidenceForRule(premise, consequence);
double tempLift = liftForRule(premise, consequence, consequenceUnconditionedCounter);
double tempLev =
leverageForRule(
premise, consequence, premise.m_counter, consequenceUnconditionedCounter);
double tempConv =
convictionForRule(
premise, consequence, premise.m_counter, consequenceUnconditionedCounter);
switch (metricType) {
case 1:
metric = tempLift;
break;
case 2:
metric = tempLev;
break;
case 3:
metric = tempConv;
break;
default:
throw new Exception("ItemSet: Unknown metric type!");
}
if (!(metric < minMetric)
&& (significanceLevel == -1 || !(chiSquared > significanceLevel))) {
premises.addElement(premise);
consequences.addElement(consequence);
conf.addElement(new Double(tempConf));
lift.addElement(new Double(tempLift));
lev.addElement(new Double(tempLev));
conv.addElement(new Double(tempConv));
}
}
}
}
rules[0] = premises;
rules[1] = consequences;
rules[2] = conf;
rules[3] = lift;
rules[4] = lev;
rules[5] = conv;
return rules;
}
