private static Tuple <List <Sequence>, int> getPrefixSpanResults(string filepath, double minSupport, bool print)
{
// current batch to process
List <Sequence> sequenceList = new List <Sequence>();
// open the CSV file
var reader = new StreamReader(File.OpenRead(filepath));
while (!reader.EndOfStream)
{
// read the current line
string line = reader.ReadLine();
// split the line by comma (assuming CSV file)
string [] transactions = line.Split(new string[] { " -1 " }, StringSplitOptions.None);
List <Transaction> tr = new List <Transaction>(transactions.Length);
for (int i = 0; i < transactions.Length - 1; i++)
{
string [] items = transactions[i].Split(' ');
Transaction trans = new Transaction(items);
tr.Add(trans);
}
Sequence sequence = new Sequence(tr);
sequenceList.Add(sequence);
}
PrefixSpan algorithm = new PrefixSpan(sequenceList);
// extract the frequent patterns
List <Sequence> frequentSequences = algorithm.MinSupportFrequentSequences(minSupport);
return(new Tuple <List <Sequence>, int>(frequentSequences, sequenceList.Count));
}
private static long processFile(string filepath, double minSupport, int maxLength)
{
// measure the time it takes for the algorithm to complete
Stopwatch stopwatch = Stopwatch.StartNew();
// current batch to process
List <Sequence> sequences = new List <Sequence>(5000000);
// read the csv file
var reader = new StreamReader(File.OpenRead(filepath));
while (!reader.EndOfStream)
{
// read the current line
string line = reader.ReadLine();
// split the line by comma (assuming CSV file)
string [] transactions = line.Split(new string[] { " -1 " }, StringSplitOptions.None);
List <Transaction> tr = new List <Transaction>(transactions.Length);
for (int i = 0; i < transactions.Length - 1; i++)
{
string [] items = transactions[i].Split(' ');
Transaction trans = new Transaction(items);
tr.Add(trans);
}
Sequence sequence = new Sequence(tr);
sequences.Add(sequence);
}
long csvReadingTime = stopwatch.ElapsedMilliseconds;
PrefixSpan algo = new PrefixSpan(sequences);
var frequentSequences = algo.MinSupportFrequentSequences(minSupport);
frequentSequences.Sort(Sequence.SequenceSorter);
frequentSequences.ForEach(Console.WriteLine);
Console.WriteLine("*****************************************************************************");
Console.WriteLine("Done with PrefixSpan results.");
Console.WriteLine("Number of frequent sequences: " + frequentSequences.Count + " sequences.");
Console.WriteLine("Minimum support: " + minSupport);
Console.WriteLine("CSV reading time: " + csvReadingTime + "ms.");
Console.WriteLine("Total run time: " + stopwatch.ElapsedMilliseconds + "ms.");
Console.WriteLine("*****************************************************************************");
return(stopwatch.ElapsedMilliseconds);
}
private static long processCSVFile(string filepath, double minSupport)
{
Stopwatch stopwatch = Stopwatch.StartNew(); //creates and start the instance of Stopwatch
// current batch to process
List <Sequence> sequenceList = new List <Sequence>();
// open the CSV file
var reader = new StreamReader(File.OpenRead(filepath));
while (!reader.EndOfStream)
{
// read the current line
string line = reader.ReadLine();
// split the line by comma (assuming CSV file)
string [] transactions = line.Split(new string[] { " -1 " }, StringSplitOptions.None);
List <Transaction> tr = new List <Transaction>(transactions.Length);
for (int i = 0; i < transactions.Length - 1; i++)
{
string [] items = transactions[i].Split(' ');
Transaction trans = new Transaction(items);
tr.Add(trans);
}
Sequence sequence = new Sequence(tr);
sequenceList.Add(sequence);
}
PrefixSpan algorithm = new PrefixSpan(sequenceList);
List <Sequence> frequentSequences = algorithm.MinSupportFrequentSequences(minSupport);
// stop the stopwatch after frequent patterns are
// returned
stopwatch.Stop();
return(stopwatch.ElapsedMilliseconds);
}
private async Task runPrefixSpan(BenchmarkRequest request, IServerStreamWriter <BenchmarkReply> responseStream, ServerCallContext context)
{
Console.WriteLine("run prefix span");
await Task.Run(async() => {
Stopwatch stopwatch = Stopwatch.StartNew(); //creates and start the instance of Stopwatch
List <Sequence> sequenceList = new List <Sequence>();
// open the CSV file
using (var reader = new StreamReader(File.OpenRead(request.File)))
{
while (!reader.EndOfStream)
{
// read the current line
string line = reader.ReadLine();
// split the line by comma (assuming CSV file)
string [] transactions = line.Split(new string[] { " -1 " }, StringSplitOptions.None);
List <Transaction> tr = new List <Transaction>(transactions.Length);
for (int i = 0; i < transactions.Length - 1; i++)
{
string [] items = transactions[i].Split(' ');
Transaction trans = new Transaction(items);
tr.Add(trans);
}
Sequence sequence = new Sequence(tr);
sequenceList.Add(sequence);
}
}
PrefixSpan algorithm = new PrefixSpan(sequenceList);
List <Sequence> frequentSequences = algorithm.MinSupportFrequentSequences(request.Support);
// stop the stopwatch after frequent patterns are
// returned
stopwatch.Stop();
var reply = new BenchmarkReply
{
NrProcessedRecords = request.DBSize,
ReplyType = ReplyType.Batch,
Iteration = 0,
Error = 0,
BatchRuntimeInMillis = stopwatch.ElapsedMilliseconds,
TotalRuntimeInMillis = stopwatch.ElapsedMilliseconds
};
var ser = JsonConvert.SerializeObject(frequentSequences, _jsonSettings);
reply.SequencesInJson = ser;
await responseStream.WriteAsync(reply);
reply = new BenchmarkReply
{
ReplyType = ReplyType.Complete,
TotalRuntimeInMillis = stopwatch.ElapsedMilliseconds
};
await responseStream.WriteAsync(reply);
});
}
