private async Task runBatchMiner(BenchmarkRequest request, IServerStreamWriter <BenchmarkReply> responseStream, ServerCallContext context)
{
Console.WriteLine("run batch miner " + request.UseTopK);
await Task.Run(async() => {
BatchMiner miner = new BatchMiner();
var batchStopwatch = new Stopwatch();
batchStopwatch.Restart();
long totalTimeElapsed = 0;
//int numBlocks = (int)Math.Ceiling(request.DBSize/(double)request.SampleSize);
await miner.ProcessCSVFile(
request.File,
request.Support,
request.K,
request.UseTopK,
request.SampleSize,
request.DBSize,
request.ErrorTolerance,
async delegate(List <Sequence> frequentSequencePatterns, double error, int iteration)
{
batchStopwatch.Stop();
totalTimeElapsed += batchStopwatch.ElapsedMilliseconds;
var reply = new BenchmarkReply
{
NrProcessedRecords = miner.NumTransactionsProcessed,
ReplyType = ReplyType.Batch,
Iteration = iteration,
Error = error,
BatchRuntimeInMillis = batchStopwatch.ElapsedMilliseconds,
TotalRuntimeInMillis = totalTimeElapsed,
PrevBlockFileReadingTime = miner.PrevBlockFileReadingTime,
PrevBlockPreProcessingRuntime = miner.Algorithm.PrevBlockPreProcessingRuntime,
PrevBlockPrefixSpanRuntime = miner.Algorithm.PrevBlockPrefixSpanRuntime,
PrevBlockSubsequenceMatchingRuntime = miner.Algorithm.PrevBlockSubsequenceMatchingRuntime
};
var ser = JsonConvert.SerializeObject(frequentSequencePatterns, _jsonSettings);
reply.SequencesInJson = ser;
await responseStream.WriteAsync(reply);
batchStopwatch.Restart();
},
long.MaxValue
);
await responseStream.WriteAsync(new BenchmarkReply
{
ReplyType = ReplyType.Complete,
TotalRuntimeInMillis = totalTimeElapsed
});
});
}
private static async Task <long> processFile(string filepath, double minSupport, int k, bool mineTopK, int sampleSize, int dbSize, long killTime)
{
BatchMiner batchPatternMiner = new BatchMiner();
batchStopwatch.Restart();
totalTimeElapsed = 0;
await batchPatternMiner.ProcessCSVFile(
filepath,
minSupport,
k,
mineTopK,
sampleSize,
dbSize,
errorTolerance,
async delegate(List <Sequence> frequentSequencePatterns, double error, int iteration)
{
onBatchResults(batchPatternMiner, frequentSequencePatterns, error, iteration);
},
killTime);
batchStopwatch.Stop();
// print out final result stats
Console.WriteLine("********************************************************************************");
Console.WriteLine("ProSecCo completed.");
Console.WriteLine("Number of frequent sequences: " + numSequences + " sequences");
if (mineTopK)
{
Console.WriteLine("Top-k: " + k);
}
else
{
Console.WriteLine("Minimum support: " + minSupport);
}
Console.WriteLine("Total runtime: " + totalTimeElapsed + "ms");
Console.WriteLine("Sum of runtimes: " + sumComponentTimes + "ms");
Console.WriteLine("********************************************************************************");
return(totalTimeElapsed);
}
private static async Task <long> processCSVFileProgressive(string filepath, double minSupport, int k, bool mineTopK, int sampleSize, int numBlocks)
{
BatchMiner miner = new BatchMiner();
batchStopwatch.Restart();
totalTimeElapsed = 0;
await miner.ProcessCSVFile(
filepath,
minSupport,
k,
mineTopK,
sampleSize,
numBlocks,
ERROR_TOLERANCE,
new FrequentSequencesBatchResult(onBatchResultsNoOutput),
long.MaxValue
);
batchStopwatch.Stop();
return(totalTimeElapsed);
}
// TESTS
private static async Task testSupportDifference(List <Sequence> expectedResults, double minSupport, int k, bool mineTopK, int sampleSize, int datasetSize)
{
batchProcessingTimes = new List <double>();
batchErrors = new List <double>();
Console.WriteLine("----------------------------------------------------");
Console.WriteLine("MIN SUPPORT: " + minSupport);
Console.WriteLine("----------------------------------------------------");
BatchMiner miner = new BatchMiner();
List <Sequence> results = new List <Sequence>();
List <double> supportDiffMax = new List <double>();
List <double> supportDiffMedian = new List <double>();
List <double> falseNegatives = new List <double>();
List <double> falsePositives = new List <double>();
int minCount = 0;
for (var i = 0; i < NUM_TRIALS; i++)
{
currentTrialNumber = i;
currentBatchNumber = 0;
batchStopwatch.Restart();
// test difference between PrefixSpan and PrefixSpan supports
miner = new BatchMiner();
await miner.ProcessCSVFile(
FILE_PATH,
minSupport,
k,
mineTopK,
sampleSize,
(int)Math.Ceiling(datasetSize / (double)sampleSize),
ERROR_TOLERANCE,
async (List <Sequence> frequentSequences, double error, int iteration) => {
await onBatchResultsNoOutput(frequentSequences, error, iteration);
results = frequentSequences;
minCount = (int)Math.Floor(datasetSize *minSupport);
Tuple <double, double, int, int> supportDiffs = getSupportDifference(expectedResults, results, datasetSize, miner.NumTransactionsProcessed, minCount);
if (i == 0)
{
batchErrors.Add(error);
supportDiffMedian.Add(supportDiffs.Item1);
supportDiffMax.Add(supportDiffs.Item2);
falsePositives.Add(supportDiffs.Item3);
falseNegatives.Add(supportDiffs.Item4);
}
else
{
batchErrors[currentBatchNumber] += (error);
supportDiffMedian[currentBatchNumber] += (supportDiffs.Item1);
supportDiffMax[currentBatchNumber] += (supportDiffs.Item2);
falsePositives[currentBatchNumber] += (supportDiffs.Item3);
falseNegatives[currentBatchNumber] += (supportDiffs.Item4);
}
currentBatchNumber++;
},
killAfter : long.MaxValue
);
}
minCount = (int)Math.Floor(miner.NumTransactionsProcessed * minSupport);
Console.WriteLine("Min support: " + minSupport);
Tuple <double, double, int, int> result = getSupportDifference(expectedResults, results, datasetSize, miner.NumTransactionsProcessed, minCount);
Console.WriteLine("Median support diff: " + result.Item1 / (double)miner.NumTransactionsProcessed);
Console.WriteLine("Max support diff: " + result.Item2 / (double)miner.NumTransactionsProcessed);
Console.WriteLine("Total Results: " + results.Count);
Console.WriteLine("False positives: " + result.Item3);
Console.WriteLine("False negatives: " + result.Item4);
Console.WriteLine("Sample size: " + sampleSize);
Console.WriteLine("----------------------------------------------------");
var divBy = (double)(NUM_TRIALS * STAT_SAMPLE_INTERVAL);
Console.Write("PROCESSING TIMES: \n[");
int index = 0;
int batch = 0;
double currentVal = 0.0;
foreach (var time in batchProcessingTimes)
{
currentVal += time;
if (batch == STAT_SAMPLE_INTERVAL)
{
Console.Write("(" + index + ", " + Math.Round(currentVal / divBy) + ")");
currentVal = 0;
index++;
batch = 0;
}
batch++;
}
Console.WriteLine("]");
Console.Write("ERRORS (median): \n[");
index = 0;
batch = 0;
currentVal = 0.0;
foreach (var median in supportDiffMedian)
{
currentVal += median;
if (batch == STAT_SAMPLE_INTERVAL)
{
Console.Write("(" + index + ", " + currentVal / divBy + ")");
index++;
currentVal = 0;
batch = 0;
}
batch++;
}
Console.WriteLine("]\n");
Console.Write("ERRORS (max): \n[");
index = 0;
batch = 0;
currentVal = 0.0;
foreach (var max in supportDiffMax)
{
currentVal += max;
if (batch == STAT_SAMPLE_INTERVAL)
{
Console.Write("(" + index + ", " + currentVal / divBy + ")");
index++;
batch = 0;
currentVal = 0;
}
batch++;
}
Console.WriteLine("]\n\n");
Console.Write("FALSE-POSITIVES: \n[");
index = 0;
batch = 0;
currentVal = 0.0;
foreach (var pos in falsePositives)
{
currentVal += pos;
if (batch == STAT_SAMPLE_INTERVAL)
{
Console.Write("(" + index + ", " + currentVal / divBy + ")");
index++;
batch = 0;
currentVal = 0;
}
batch++;
}
Console.WriteLine("]\n\n");
Console.Write("FALSE-NEGATIVES: \n[");
index = 0;
batch = 0;
currentVal = 0.0;
foreach (var neg in falseNegatives)
{
currentVal += neg;
if (batch == STAT_SAMPLE_INTERVAL)
{
Console.Write("(" + index + ", " + currentVal / divBy + ")");
index++;
batch = 0;
currentVal = 0;
}
batch++;
}
Console.WriteLine("]\n\n");
}