protected override Itemset[] ScanDbAndCountCandidatesSupport(Itemset[] candidates)
{
foreach (var transaction in _db.Transactions)
{
if (IsReducted(transaction))
{
continue;
}
bool containsCandidates = false;
foreach (var candidate in candidates)
{
if (ItemsetHelper.IsSubset(transaction.Items, candidate.Items))
{
candidate.Support++;
containsCandidates = true;
}
}
if (containsCandidates)
{
continue;
}
ReduceTransaction(transaction);
}
return(candidates);
}
protected virtual Itemset[] PruneCandidatesByAprioriSlim(UInt32[][] candidatesItemset, Itemset[] frequentItemset)
{
var frequentCandidates = new List <Itemset>();
var frequent = new HashSet <UInt32[]>(frequentItemset.Select(i => i.Items), new ItemsEqualityComparer());
for (int i = 0; i < candidatesItemset.Length; i++)
{
bool hasAllSubItemsetFrequent = true;
// TODO: Lazy generating, we can stop generating subitemset if we know that an one is not frequent.
List <UInt32[]> subItemsets = ItemsetHelper.GenerateOneLevelSubItemsets(candidatesItemset[i]);
for (int j = 0; j < subItemsets.Count; j++)
{
if (frequent.Contains(subItemsets[j]))
{
continue;
}
hasAllSubItemsetFrequent = false;
break;
}
if (hasAllSubItemsetFrequent)
{
frequentCandidates.Add(new Itemset(candidatesItemset[i]));
}
}
return(frequentCandidates.ToArray());
}
protected override Itemset[] ScanDbAndCountCandidatesSupport(Itemset[] candidates)
{
if (candidates.Length == 0)
{
return(candidates);
}
Dictionary <UInt32[], Itemset> candidatesDictionary = CreateCandidatesDictionary(candidates);
foreach (var transaction in _db.Transactions)
{
List <UInt32[]> transactionsSubItemsets = ItemsetHelper.GenerateSubItemsets(transaction.Items, candidates[0].Length);
for (int i = 0; i < transactionsSubItemsets.Count; i++)
{
Itemset itemset;
if (candidatesDictionary.TryGetValue(transactionsSubItemsets[i], out itemset))
{
itemset.Support++;
}
}
}
return(candidates);
}
private void CountSupportForCandidatesInRange(Tuple <int, int> range)
{
for (int i = range.Item1; i < range.Item2; i++)
{
if (ItemsetHelper.IsSubset(_actualTransaction.Items, _candidates[i].Items))
{
_candidates[i].Support++;
}
}
}
public void Can_Generate_SubItemset_From_2_Length_Itemset()
{
Itemset itemset = new Itemset(new uint[] { 0, 1 });
Itemset[] result = ItemsetHelper.GenerateOneLevelSubItemsets(itemset).ToArray();
Assert.AreEqual(2, result.Length);
Assert.AreEqual(1, result[0].Length);
Assert.AreEqual(1, result[1].Length);
Assert.AreEqual(result[0][0], 0);
Assert.AreEqual(result[1][0], 1);
}
protected virtual Itemset[] ScanDbAndCountCandidatesSupport(Itemset[] candidates)
{
foreach (var transaction in _db.Transactions)
{
foreach (var candidate in candidates)
{
if (ItemsetHelper.IsSubset(transaction.Items, candidate.Items))
{
candidate.Support++;
}
}
}
return(candidates);
}
public void Can_Calculate_SubItemset()
{
bool result = ItemsetHelper.IsSubset(Itemset.Create(1), Itemset.Create(2));
Assert.AreEqual(false, result);
result = ItemsetHelper.IsSubset(Itemset.Create(1, 2), Itemset.Create(2));
Assert.AreEqual(true, result);
result = ItemsetHelper.IsSubset(Itemset.Create(1, 2, 3), Itemset.Create(2));
Assert.AreEqual(true, result);
result = ItemsetHelper.IsSubset(Itemset.Create(1, 2, 3), Itemset.Create(2, 3));
Assert.AreEqual(true, result);
result = ItemsetHelper.IsSubset(Itemset.Create(1, 2, 3), Itemset.Create(2, 3, 4));
Assert.AreEqual(false, result);
}
public void Can_Generate_All_3_Length_SubItemsets()
{
Itemset itemset = new Itemset(new uint[] { 1, 2, 3, 5, 6 });
var result = ItemsetHelper.GenerateSubItemsets(itemset.Items, 3);
Assert.AreEqual(10, result.Count);
// Length
Assert.AreEqual(3, result[0].Length);
Assert.AreEqual(3, result[9].Length);
// SubItemsets
Assert.AreEqual(result[0][0], 1);
Assert.AreEqual(result[0][1], 2);
Assert.AreEqual(result[0][2], 3);
Assert.AreEqual(result[9][0], 3);
Assert.AreEqual(result[9][1], 5);
Assert.AreEqual(result[9][2], 6);
}
public void Can_Enumerate_All_3_Length_SubItemsets()
{
Itemset itemset = new Itemset(new uint[] { 1, 2, 3, 5, 6 });
UInt32[][] result = ItemsetHelper.EnumerateSubItemsets(itemset.Items, 3).ToArray();
Assert.AreEqual(10, result.Length);
// Length
Assert.AreEqual(3, result[0].Length);
Assert.AreEqual(3, result[9].Length);
// SubItemsets
Assert.AreEqual(result[0][0], 1);
Assert.AreEqual(result[0][1], 2);
Assert.AreEqual(result[0][2], 3);
Assert.AreEqual(result[9][0], 3);
Assert.AreEqual(result[9][1], 5);
Assert.AreEqual(result[9][2], 6);
}
public void Can_Generate_SubItemset_From_3_Length_Itemset()
{
Itemset itemset = Itemset.Create(0, 1, 2);
Itemset[] result = ItemsetHelper.GenerateOneLevelSubItemsets(itemset).ToArray();
Assert.AreEqual(3, result.Length);
Assert.AreEqual(2, result[0].Length);
Assert.AreEqual(2, result[1].Length);
Assert.AreEqual(2, result[2].Length);
Assert.AreEqual(result[0][0], 0);
Assert.AreEqual(result[0][1], 1);
Assert.AreEqual(result[1][0], 0);
Assert.AreEqual(result[1][1], 2);
Assert.AreEqual(result[2][0], 1);
Assert.AreEqual(result[2][1], 2);
}
