private double[] ExtractData(string name, DataSetPairsType type, DataSetPairsCollection dspc)
{
foreach (DataSetPairs pairs in dspc)
{
if (pairs.Name == name)
{
if (pairs.Name == "VSM")
{
pairs.Name = "VSM";
}
switch (type)
{
case DataSetPairsType.Precision:
return(ExtractDataArray(pairs.PrecisionData));
case DataSetPairsType.Recall:
return(ExtractDataArray(pairs.RecallData));
case DataSetPairsType.AveragePrecision:
return(ExtractDataArray(pairs.AveragePrecisionData));
default:
throw new ArgumentException("Unknown DataSetPairsType: " + type);
}
}
}
throw new ArgumentException("Model not found - " + name);
}
public override void Compute()
{
DataSetPairsCollection dspc = (DataSetPairsCollection)Workspace.Load("DataSetPairsCollection");
List <DatasetResults> listOfDatasetResults = new List <DatasetResults>();
List <Metric> ds1 = new List <Metric>();
// tab 1 - base precision
List <SummaryData> ds1_tab1 = new List <SummaryData>();
ds1_tab1.Add(new SummaryData("VSM", ExtractData("VSM", DataSetPairsType.AveragePrecision, dspc)));
ds1_tab1.Add(new SummaryData("JS", ExtractData("JS", DataSetPairsType.AveragePrecision, dspc)));
ds1_tab1.Add(new SummaryData("RTM", ExtractData("RTM", DataSetPairsType.AveragePrecision, dspc)));
ds1.Add(new AveragePrecisionMetric(ds1_tab1, 0.0, ""));//, "Base - Precision"));
// tab 2 - combined precision
List <SummaryData> ds1_tab2 = new List <SummaryData>();
ds1_tab2.Add(new SummaryData("VSM_JS", ExtractData("VSM_JS", DataSetPairsType.AveragePrecision, dspc)));
ds1_tab2.Add(new SummaryData("VSM_RTM", ExtractData("VSM_RTM", DataSetPairsType.AveragePrecision, dspc)));
ds1_tab2.Add(new SummaryData("JS_RTM", ExtractData("JS_RTM", DataSetPairsType.AveragePrecision, dspc)));
ds1.Add(new AveragePrecisionMetric(ds1_tab2, 0.0, ""));//, "Combined - Precision"));
// tab 3 - combined PCA precision
List <SummaryData> ds1_tab3 = new List <SummaryData>();
ds1_tab3.Add(new SummaryData("VSM_JS_PCA", ExtractData("VSM_JS_PCA", DataSetPairsType.AveragePrecision, dspc)));
ds1_tab3.Add(new SummaryData("VSM_RTM_PCA", ExtractData("VSM_RTM_PCA", DataSetPairsType.AveragePrecision, dspc)));
ds1_tab3.Add(new SummaryData("JS_RTM_PCA", ExtractData("JS_RTM_PCA", DataSetPairsType.AveragePrecision, dspc)));
ds1.Add(new AveragePrecisionMetric(ds1_tab3, 0.0, ""));//, "Combined (PCA) - Precision"));
listOfDatasetResults.Add(new DatasetResults("Dataset", ds1));
Workspace.Store("ListOfDatasetResults", listOfDatasetResults);
}
private double[] ExtractData(string name, DataSetPairsType type, DataSetPairsCollection dspc)
{
foreach (DataSetPairs pairs in dspc)
{
if (pairs.Name == name)
{
if (pairs.Name == "VSM")
{
pairs.Name = "VSM";
}
switch (type)
{
case DataSetPairsType.Precision:
return ExtractDataArray(pairs.PrecisionData);
case DataSetPairsType.Recall:
return ExtractDataArray(pairs.RecallData);
case DataSetPairsType.AveragePrecision:
return ExtractDataArray(pairs.AveragePrecisionData);
default:
throw new ArgumentException("Unknown DataSetPairsType: " + type);
}
}
}
throw new ArgumentException("Model not found - " + name);
}
public static DataSetPairsCollection ComputeAll(TLSimilarityMatrix sims, TLSimilarityMatrix oracle)
{
DataSetPairsCollection dspc = new DataSetPairsCollection();
foreach (RecallLevel level in Enum.GetValues(typeof(RecallLevel)))
{
dspc.Add(Compute(sims, oracle, level));
}
return dspc;
}
public static DataSetPairsCollection ComputeAll(TLSimilarityMatrix sims, TLSimilarityMatrix oracle)
{
DataSetPairsCollection dspc = new DataSetPairsCollection();
foreach (RecallLevel level in Enum.GetValues(typeof(RecallLevel)))
{
dspc.Add(Compute(sims, oracle, level));
}
return(dspc);
}
private static void MyMetrics(ref Info info, string output, TLSimilarityMatrix matrix, TLSimilarityMatrix oracle)
{
DataSetPairsCollection metrics = OverallMetricsComputation.ComputeAll(matrix, oracle);
TextWriter dataFile = File.CreateText(output);
for (int i = 0, j = 10; i < metrics.Count; i++, j += 10)
{
dataFile.WriteLine("{0} {1}", j, metrics[i].PrecisionData[0].Value);
}
dataFile.Flush();
dataFile.Close();
}
public static void Run(ref Info info)
{
TLSimilarityMatrix vsm = Similarities.Import(info.ResultsDirectory + @"\CSMRTools\RoccoResults\EasyClinic UC-CC.VSM.sims");
TLSimilarityMatrix oracle = Oracle.Import(info.ResultsDirectory + @"\CSMRTools\RoccoResults\EasyClinic UC-CC.oracle");
TLSimilarityMatrix usage = Oracle.Import(info.ResultsDirectory + @"\CSMRTools\RoccoResults\EasyClinic UC-CC.relationships");
TLSimilarityMatrix udcsti = UDCSTI.Compute(vsm, usage, oracle);
Similarities.Export(udcsti, info.ResultsDirectory + @"\CSMRTools\RoccoResults\EasyClinic UC-CC.VSM_UDCSTI.MYsims");
DataSetPairsCollection metrics = OverallMetricsComputation.ComputeAll(udcsti, oracle);
TextWriter dataFile = File.CreateText(info.ResultsDirectory + @"\CSMRTools\RoccoResults\EasyClinic UC-CC.VSM_UDCSTI.MYsims.metrics");
for (int i = 0, j = 10; i < metrics.Count; i++, j += 10)
{
dataFile.WriteLine("{0} {1}", j, metrics[i].PrecisionData[0].Value);
}
dataFile.Flush();
dataFile.Close();
}
public static DataSetPairs Compute(TLSimilarityMatrix sims, TLSimilarityMatrix oracle, RecallLevel recall)
{
TLSimilarityMatrix matrix = Similarities.CreateMatrix(MetricsUtil.GetLinksAtRecall(sims, oracle, recall));
matrix.Threshold = double.MinValue;
DataSetPairs pairs = new DataSetPairs();
foreach (string sourceArtifact in oracle.SourceArtifactsIds)
{
TLLinksList links = matrix.GetLinksAboveThresholdForSourceArtifact(sourceArtifact);
links.Sort();
int totalCorrect = oracle.GetLinksAboveThresholdForSourceArtifact(sourceArtifact).Count;
int numCorrect = 0;
int totalRead = 0;
double totalAvgPrecision = 0.0;
foreach (TLSingleLink link in links)
{
totalRead++;
if (oracle.IsLinkAboveThreshold(link.SourceArtifactId, link.TargetArtifactId))
{
numCorrect++;
totalAvgPrecision += numCorrect / (double)totalRead;
}
}
pairs.PrecisionData.Add(new KeyValuePair <string, double>(sourceArtifact, numCorrect / Convert.ToDouble(links.Count)));
pairs.RecallData.Add(new KeyValuePair <string, double>(sourceArtifact, Convert.ToDouble(numCorrect) / totalCorrect));
pairs.AveragePrecisionData.Add(new KeyValuePair <string, double>(sourceArtifact, totalAvgPrecision / totalCorrect));
}
pairs.MeanAveragePrecisionData.Add(new KeyValuePair <string, double>("#TOTAL", DataSetPairsCollection.CalculateAverage(pairs.AveragePrecisionData)));
return(pairs);
}
