/**
* Returns an array containing the number of process instances that are considered to be fast,
* i.e. have a low throughput time (place 0 in array), the number of process instances that are
* slow (place 1 in array) and the number of process instances that are considered to be of normal
* speed (place 2 in array). Based on fastestPercentage, slowestPercentage and timeList (thus
* method calculateProcessMetrics() should be called before this one)
*
* @param fastestPercentage double: the percentage of measurements that is to be counted as fast
* @param slowestPercentage double: the percentage of measurements that is to be counted as slow
* @return int[]
*/
public int[] getSizes(double fastestPercentage, double slowestPercentage) {
int[] sizes = new int[3];
String sizeString;
int length = timeStats.getValues().length;
sizeString = Math.round((length * fastestPercentage) / 100.0) + "";
sizes[0] = Integer.parseInt(sizeString);
if (sizes[0] != length) {
sizeString = Math.round((length * slowestPercentage) / 100.0) + "";
sizes[1] = Integer.parseInt(sizeString);
if ((sizes[0] + sizes[1]) > length) {
// Make sure that sizes[0] + sizes[1] remains smaller than
// the number of measurements in timeList (rounding could mess
// this up)
sizes[1] = length - sizes[0];
}
} else {
sizes[1] = 0;
}
sizes[2] = length - sizes[0] - sizes[1];
return sizes;
}
private void loadSociogramData(Collection<VertexRecord> records, SQLDumpReader sqlData) {
logger.info("Loading sociogram data...");
Map<String, VertexRecord> map = sqlData.getFullAlterKeyMappping(records);
TObjectIntHashMap<Vertex> rawDegrees = new TObjectIntHashMap<Vertex>();
for (Vertex v : proj.getVertices()) {
rawDegrees.put(v, v.getNeighbours().size());
}
int edgecnt = 0;
int doublecnt = 0;
int egoEdge = 0;
Set<Vertex> notOkVertices = new HashSet<Vertex>();
Set<Vertex> okVertices = new HashSet<Vertex>();
DescriptiveStatistics notOkStats = new DescriptiveStatistics();
DescriptiveStatistics okStats = new DescriptiveStatistics();
DescriptiveStatistics numDistr = new DescriptiveStatistics();
DescriptiveStatistics numDistrNoZero = new DescriptiveStatistics();
DescriptiveStatistics sizeDistr = new DescriptiveStatistics();
TDoubleArrayList sizeValues = new TDoubleArrayList();
TDoubleArrayList kSizeValues = new TDoubleArrayList();
TDoubleArrayList numValues = new TDoubleArrayList();
TDoubleArrayList numValues2 = new TDoubleArrayList();
TDoubleArrayList kNumValues = new TDoubleArrayList();
for (VertexRecord record : records) {
if (record.isEgo) {
List<Set<String>> cliques = sqlData.getCliques(record);
numDistr.addValue(cliques.size());
Vertex v = idMap.get(record.id);
numValues.add(cliques.size());
kNumValues.add(v.getNeighbours().size());
if (!cliques.isEmpty()) numDistrNoZero.addValue(cliques.size());
for (Set<String> clique : cliques) {
sizeDistr.addValue(clique.size());
sizeValues.add(clique.size());
kSizeValues.add(rawDegrees.get(projMap.get(v)));
numValues2.add(cliques.size());
List<SocialSparseVertex> vertices = new ArrayList<SocialSparseVertex>(clique.size());
for (String alter : clique) {
VertexRecord r = map.get(record.egoSQLId + alter);
if (r != null) {
SocialSparseVertex vertex = idMap.get(r.id);
if (vertex != null) {
vertices.add(vertex);
} else {
logger.warn("Vertex not found.");
}
} else {
logger.warn("Record not found.");
}
}
for (int i = 0; i < vertices.size(); i++) {
for (int j = i + 1; j < vertices.size(); j++) {
SampledVertexDecorator<SocialSparseVertex> vProj1 = projMap.get(vertices.get(i));
SampledVertexDecorator<SocialSparseVertex> vProj2 = projMap.get(vertices.get(j));
if (!vProj1.isSampled() && !vProj2.isSampled()) {
if (Math.random() < 0.62) {
SocialSparseEdge socialEdge =
builder.addEdge(graph, vertices.get(i), vertices.get(j));
if (socialEdge != null) {
projBuilder.addEdge(proj, vProj1, vProj2, socialEdge);
edgecnt++;
if (vProj1.isSampled() || vProj2.isSampled()) {
egoEdge++;
if (vProj1.isSampled()) notOkVertices.add(vProj1);
else notOkVertices.add(vProj2);
}
} else {
doublecnt++;
if (vProj1.isSampled()) okVertices.add(vProj1);
else if (vProj2.isSampled()) okVertices.add(vProj2);
}
}
}
}
}
}
}
}
for (Vertex v : okVertices) okStats.addValue(rawDegrees.get(v));
for (Vertex v : notOkVertices) notOkStats.addValue(rawDegrees.get(v));
try {
TDoubleDoubleHashMap hist =
Histogram.createHistogram(okStats, new LinearDiscretizer(1), false);
StatsWriter.writeHistogram(
hist,
"k",
"n",
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/k_ok.txt");
TDoubleDoubleHashMap hist2 =
Histogram.createHistogram(notOkStats, new LinearDiscretizer(1), false);
StatsWriter.writeHistogram(
hist2,
"k",
"n",
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/k_notok.txt");
TDoubleDoubleHashMap ratio = new TDoubleDoubleHashMap();
double[] keys = hist.keys();
for (double k : keys) {
double val1 = hist2.get(k);
double val2 = hist.get(k);
ratio.put(k, val1 / (val2 + val1));
}
StatsWriter.writeHistogram(
ratio,
"k",
"p",
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/k_ratio.txt");
logger.info("Mean num of cliques: " + numDistrNoZero.getMean());
logger.info("Mean size: " + sizeDistr.getMean());
logger.info("Median num of cliques: " + StatUtils.percentile(numDistrNoZero.getValues(), 50));
logger.info("Median size: " + StatUtils.percentile(sizeDistr.getValues(), 50));
TDoubleDoubleHashMap histNum =
Histogram.createHistogram(
numDistrNoZero,
FixedSampleSizeDiscretizer.create(numDistrNoZero.getValues(), 2, 20),
true);
Histogram.normalize(histNum);
StatsWriter.writeHistogram(
histNum,
"num",
"freq",
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/numCliques.txt");
TDoubleDoubleHashMap histSize =
Histogram.createHistogram(
sizeDistr, FixedSampleSizeDiscretizer.create(sizeDistr.getValues(), 2, 20), true);
Histogram.normalize(histSize);
StatsWriter.writeHistogram(
histSize,
"size",
"freq",
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/numPersons.txt");
Discretizer discretizer =
FixedSampleSizeDiscretizer.create(kSizeValues.toNativeArray(), 20, 20);
TDoubleArrayList valuesX = new TDoubleArrayList();
for (int i = 0; i < kSizeValues.size(); i++) {
valuesX.add(discretizer.discretize(kSizeValues.get(i)));
}
Correlations.writeToFile(
Correlations.mean(valuesX.toNativeArray(), sizeValues.toNativeArray()),
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/size_k.txt",
"k",
"size");
discretizer = FixedSampleSizeDiscretizer.create(kNumValues.toNativeArray(), 20, 20);
valuesX = new TDoubleArrayList();
for (int i = 0; i < kNumValues.size(); i++) {
valuesX.add(discretizer.discretize(kNumValues.get(i)));
}
Correlations.writeToFile(
Correlations.mean(valuesX.toNativeArray(), numValues.toNativeArray()),
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/num_k.txt",
"k",
"n");
Correlations.writeToFile(
Correlations.mean(numValues2.toNativeArray(), sizeValues.toNativeArray()),
"/Users/jillenberger/Work/socialnets/data/ivt2009/11-2011/augmented/size_num.txt",
"num",
"size");
} catch (FileNotFoundException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
logger.info(
String.format("Inserted %1$s edges, %2$s edges already present.", edgecnt, doublecnt));
logger.info(String.format("Inserted %1$s edges between at least one ego.", egoEdge));
}