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());
}
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);
}
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);
}
