public List <SequenceSupport> FindSequentialPatterns(List <List <List <string> > > datasetSequences, int support)
{
List <SequenceSupport> frequentSequences = new List <SequenceSupport>();
List <SequenceSupport> candidates = new List <SequenceSupport>();
float supportFrequency = ((float)support / 100) * datasetSequences.Count();
//Find all 1 items
List <string> items = new List <string>();
items = datasetSequences.SelectMany(a => a.SelectMany(i => i)).Distinct().OrderBy(a => a).ToList();
foreach (string item in items)
{
SequenceSupport candidate = new SequenceSupport();
candidate.sequence = new List <List <string> >()
{
new List <string>()
{
item
}
};
candidates.Add(candidate);
}
//Calculate 1 itemsets frequency
candidates = CountSupport(datasetSequences, candidates);
//Find frequent 1 itemsset
frequentSequences = candidates.Where(a => a.support >= supportFrequency).ToList();
List <List <List <string> > > lastFrequentSequences = frequentSequences.Select(a => a.sequence).ToList();
int sequenceLenght = 0;
System.Diagnostics.Debug.WriteLine("Start Sequence Lenght:" + sequenceLenght++ + " : " + DateTime.Now.TimeOfDay);
bool l1Join = true;
while (lastFrequentSequences.Count > 0)
{
System.Diagnostics.Debug.WriteLine("Start Sequence Lenght:" + sequenceLenght++ + " : " + DateTime.Now.TimeOfDay);
if (l1Join == true)
{
candidates = CandidateGenerationC2(lastFrequentSequences);
l1Join = false;
}
else
{
candidates = CandidateGeneration(lastFrequentSequences);
candidates = CandidatePruning(candidates, lastFrequentSequences);
}
candidates = SupportCounting(datasetSequences, candidates);
candidates = CandidateElimination(candidates, supportFrequency);
frequentSequences.AddRange(candidates);
lastFrequentSequences = candidates.Select(a => a.sequence).ToList();
}
//comment!
return(frequentSequences);
}
private bool candidateNotExist(List <SequenceSupport> candidates, SequenceSupport candidate)
{
bool notexist = true;
foreach (SequenceSupport existingCandidate in candidates)
{
if (CompareListofLists(existingCandidate.sequence, candidate.sequence))
{
notexist = false;
}
}
return(notexist);
}
private List <SequenceSupport> CandidateGenerationC2(List <List <List <string> > > lastFrequentSequences)
{
List <SequenceSupport> candidates = new List <SequenceSupport>();
foreach (List <List <string> > lastFrequentSequence1 in lastFrequentSequences)
{
foreach (List <List <string> > lastFrequentSequence2 in lastFrequentSequences)
{
string item1 = lastFrequentSequence1.First()[0];
string item2 = lastFrequentSequence2.First()[0];
SequenceSupport candidate1 = new SequenceSupport();
candidate1.sequence = new List <List <string> >()
{
new List <string>()
{
item1
}, new List <string>()
{
item2
}
};
candidates.Add(candidate1);
if (string.Compare(item1, item2) < 0)
{
SequenceSupport candidate2 = new SequenceSupport();
candidate2.sequence = new List <List <string> >()
{
new List <string>()
{
item1, item2
}
};
candidates.Add(candidate2);
}
}
}
return(candidates);
}
private List <SequenceSupport> ReadFrequentSequences()
{
List <SequenceSupport> sequenceSupports = new List <SequenceSupport>();
var a = _dbContext.FrequentSequentialPattern.Where(r => r.Train == true);
foreach (var item in a)
{
char[] splitChars = new[] { '(', ')', '<', '>' };
string[] itemsets = item.Sequence.Split(splitChars, StringSplitOptions.RemoveEmptyEntries);
List <List <string> > itemsetsList = new List <List <string> >();
foreach (string i in itemsets)
{
splitChars = new[] { ' ', '-' };
string[] itemset = i.Split(splitChars, StringSplitOptions.RemoveEmptyEntries);
List <string> itemsetList = new List <string>(itemset);
itemsetsList.Add(itemsetList);
}
SequenceSupport sequenceSupport = new SequenceSupport();
sequenceSupport.support = item.Support;
sequenceSupport.sequence = itemsetsList;
sequenceSupports.Add(sequenceSupport);
}
return(sequenceSupports);
}
private List <SequenceSupport> CandidateGeneration(List <List <List <string> > > lastFrequentSequences)
{
List <SequenceTrimed> SequencesTrimed = new List <SequenceTrimed>();
foreach (List <List <string> > sequence in lastFrequentSequences)
{
SequenceTrimed SequenceTrimed = new SequenceTrimed();
SequenceTrimed.sequenceOriginal = new List <List <string> >(sequence);
List <List <string> > sequenceCopied = sequence.ConvertAll(a => new List <string>(a.ToList()));
//First itemset lenght ==1 E.g. A(BC)
if (sequence[0].Count == 1)
{
sequenceCopied.RemoveAt(0);
SequenceTrimed.sequenceRemovedFirst = sequenceCopied.ConvertAll(a => new List <string>(a.ToList()));
sequenceCopied = sequence.ConvertAll(a => new List <string>(a.ToList()));
int sequenceCount = sequence.Count();
if (sequence[sequenceCount - 1].Count == 1)
{
//First itemset lenght ==1 & last itemset==1 ----- E.g. AB
sequenceCopied.RemoveAt(sequenceCount - 1);
SequenceTrimed.sequenceRemovedLast = sequenceCopied.ConvertAll(a => new List <string>(a.ToList()));
SequenceTrimed.lastItem = sequence[sequenceCount - 1].First();
SequencesTrimed.Add(SequenceTrimed);
}
else
{
//First itemset lenght == 1 & last itemset > 1 ----- E.g. A(BC)
foreach (string item in sequence[sequenceCount - 1])
{
sequenceCopied = sequence.ConvertAll(a => new List <string>(a.ToList()));
sequenceCopied[sequenceCount - 1].Remove(item);
SequenceTrimed.sequenceRemovedLast = sequenceCopied.ConvertAll(a => new List <string>(a.ToList()));
SequenceTrimed.lastItem = item;
SequencesTrimed.Add(SequenceTrimed);
}
}
}
else
{
foreach (string item in sequence[0])
{
sequenceCopied = sequence.ConvertAll(a => new List <string>(a.ToList()));
sequenceCopied[0].Remove(item);
SequenceTrimed.sequenceRemovedFirst = sequenceCopied.ConvertAll(a => new List <string>(a.ToList()));
int sequenceCount = sequence.Count();
if (sequence[sequenceCount - 1].Count == 1)
{
//First itemset lenght > 1 & last itemset == 1 ----- E.g. (AB)C
sequenceCopied.RemoveAt(sequenceCount - 1);
SequenceTrimed.sequenceRemovedLast = sequenceCopied.ConvertAll(a => new List <string>(a.ToList()));
SequenceTrimed.lastItem = sequence[sequenceCount - 1].First();
SequencesTrimed.Add(SequenceTrimed);
}
else
{
foreach (string itemlast in sequence[sequenceCount - 1])
{
//First itemset lenght > 1 & last itemset > 1 ----- E.g. (AB)(CD)
sequenceCopied = sequence.ConvertAll(a => new List <string>(a.ToList()));
sequenceCopied[sequenceCount - 1].Remove(itemlast);
SequenceTrimed.sequenceRemovedLast = sequenceCopied.ConvertAll(a => new List <string>(a.ToList()));
SequenceTrimed.lastItem = itemlast;
SequencesTrimed.Add(SequenceTrimed);
}
}
}
}
}
List <SequenceSupport> candidates = new List <SequenceSupport>();
foreach (SequenceTrimed sequenceTrimed1 in SequencesTrimed)
{
foreach (SequenceTrimed sequenceTrimed2 in SequencesTrimed)
{
if (CompareListofLists(sequenceTrimed1.sequenceRemovedFirst, sequenceTrimed2.sequenceRemovedLast))
{
SequenceSupport candidate = new SequenceSupport();
if (sequenceTrimed2.sequenceOriginal.Last().Count() == 1)
{
candidate.sequence = sequenceTrimed1.sequenceOriginal.ConvertAll(a => new List <string>(a.ToList()));
candidate.sequence.Add(new List <string>()
{
sequenceTrimed2.lastItem
});
if (candidateNotExist(candidates, candidate))
{
candidates.Add(candidate);
}
}
else
{
if (!sequenceTrimed1.sequenceOriginal.Last().Contains(sequenceTrimed2.lastItem))
{
candidate.sequence = sequenceTrimed1.sequenceOriginal.ConvertAll(a => new List <string>(a.ToList()));;
candidate.sequence.Last().Add(sequenceTrimed2.lastItem);
if (candidateNotExist(candidates, candidate))
{
candidates.Add(candidate);
}
}
}
}
}
}
return(candidates);
}