public void HACClusterDataTest4D()
{
HAC hac = new HAC();
List <Cluster> testOutput1 = hac.ClusterData(_testInput2, "average", "euclidean", 0);
List <Cluster> testOutput2 = hac.ClusterData(_testInput2, "minimum", "euclidean2", 0);
List <Cluster> testOutput3 = hac.ClusterData(_testInput2, "minimum", "manhattan", 0);
List <Cluster> testOutput4 = hac.ClusterData(_testInput2, "maximum", "euclidean", 0);
List <Cluster> testOutput5 = hac.ClusterData(_testInput2, "maximum", "manhattan", 0);
bool allEqual = true;
if (testOutput1.Count != testOutput2.Count || testOutput1.Count != testOutput3.Count ||
testOutput1.Count != testOutput4.Count || testOutput1.Count != testOutput5.Count)
{
allEqual = false;
}
else
{
foreach (Cluster c in testOutput1)
{
if (!testOutput2.Contains(c) || !testOutput3.Contains(c) ||
!testOutput4.Contains(c) || !testOutput5.Contains(c))
{
allEqual = false;
}
}
}
Assert.IsTrue(allEqual, "Outputs are not equal");
Assert.IsTrue(testOutput1.Count.Equals(_testInput2.Count * 2 - 1), "HAC output incomplete");
Assert.IsTrue(testOutput1.Contains(new Cluster(_p7)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput1.Contains(new Cluster(_p8)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput1.Contains(new Cluster(_p9)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput1.Contains(new Cluster(_p10)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput1.Contains(new Cluster(new List <Point> {
_p7, _p8
})), "HAC output 1 incomplete");
Assert.IsTrue(testOutput1.Contains(new Cluster(new List <Point> {
_p7, _p8, _p9
})), "HAC output 1 incomplete");
Assert.IsTrue(testOutput1.Contains(new Cluster(new List <Point> {
_p7, _p8, _p9, _p10
})), "HAC output 1 incomplete");
Assert.IsFalse(testOutput1.Contains(new Cluster(new List <Point> {
_p7, _p9
})), "HAC output 1 has incorrect data");
Assert.IsFalse(testOutput1.Contains(new Cluster(new List <Point> {
_p9, _p10
})), "HAC output 1 has incorrect data");
}
static void RunBenchmark(int[] pointCounts, int[] pointDims, int averageOf, string linkageMethod = "average", string distanceMethod = "euclideansquared", bool createFiles = false, string resultFilename = "benchmark_results.txt")
{
string startupMessage = $"Clustering benchmark with parameters: averageOf={averageOf}," +
$" linkageMethod=\"{linkageMethod}\", distanceMethod=\"{distanceMethod}\"";
Console.WriteLine(startupMessage);
DataIO dio = new DataIO();
HAC hac = new HAC();
using (StreamWriter file = new StreamWriter(resultFilename))
{
file.WriteLine(startupMessage);
foreach (int count in pointCounts)
{
foreach (int dim in pointDims)
{
long result = 0;
for (int i = 0; i < averageOf; ++i)
{
string inputFilename = createFiles ? $"testGenerated-{count}-{dim}-{i}.txt" : "";
string outputFilename = createFiles ? $"testOutput-{count}-{dim}-{i}.txt" : "";
Console.WriteLine($"Now clustering: {count}-{dim}-{i}...");
Stopwatch watch = new Stopwatch();
List <Point> input = dio.GenerateData(-1000000, 1000000, count, dim, inputFilename);
watch.Start();
List <Cluster> output = hac.ClusterData(input, "average", "euclidean2", 0);
watch.Stop();
result += watch.ElapsedMilliseconds;
if (createFiles)
{
dio.SaveData(output, outputFilename);
}
}
result /= averageOf;
file.WriteLine($"{count} points, {dim} dimensions = {result} milliseconds");
}
}
}
Console.WriteLine("Benchmark finished successfully");
}
public void HACClusterDataTest2D()
{
HAC hac = new HAC();
List <Cluster> testOutput = hac.ClusterData(_testInput, "average", "euclidean2", 0);
List <Cluster> testOutput2 = hac.ClusterData(_testInput, "average", "manhattan", 0);
List <Cluster> testOutput3 = hac.ClusterData(_testInput, "minimum", "euclidean", 0);
List <Cluster> testOutput4 = hac.ClusterData(_testInput, "maximum", "euclidean2", 0);
bool equal124 = true;
if (testOutput.Count != testOutput2.Count || testOutput.Count != testOutput4.Count)
{
equal124 = false;
}
else
{
foreach (Cluster c in testOutput)
{
if (!testOutput2.Contains(c) || !testOutput4.Contains(c))
{
equal124 = false;
}
}
}
Assert.IsTrue(equal124, "Outputs 1, 2 and 4 are not equal");
Assert.IsTrue(testOutput.Count.Equals(_testInput.Count * 2 - 1), "HAC output incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(_p1)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(_p2)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(_p3)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(_p4)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(_p5)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(_p6)), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(new List <Point> {
_p1, _p2
})), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(new List <Point> {
_p3, _p4
})), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(new List <Point> {
_p5, _p6
})), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(new List <Point> {
_p3, _p4, _p5, _p6
})), "HAC output 1 incomplete");
Assert.IsTrue(testOutput.Contains(new Cluster(new List <Point> {
_p1, _p2, _p3, _p4, _p5, _p6
})), "HAC output 1 incomplete");
Assert.IsFalse(testOutput.Contains(new Cluster(new List <Point> {
_p1, _p2, _p3
})), "HAC output 1 has incorrect data");
Assert.IsFalse(testOutput.Contains(new Cluster(new List <Point> {
_p1, _p6
})), "HAC output 1 has incorrect data");
Assert.IsTrue(testOutput3.Count.Equals(_testInput.Count * 2 - 1), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(_p1)), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(_p2)), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(_p3)), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(_p4)), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(_p5)), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(_p6)), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(new List <Point> {
_p1, _p2
})), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(new List <Point> {
_p3, _p4
})), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(new List <Point> {
_p5, _p6
})), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(new List <Point> {
_p1, _p2, _p3, _p4
})), "HAC output 3 incomplete");
Assert.IsTrue(testOutput3.Contains(new Cluster(new List <Point> {
_p1, _p2, _p3, _p4, _p5, _p6
})), "HAC output 3 incomplete");
Assert.IsFalse(testOutput3.Contains(new Cluster(new List <Point> {
_p1, _p2, _p3
})), "HAC output 3 has incorrect data");
Assert.IsFalse(testOutput3.Contains(new Cluster(new List <Point> {
_p1, _p6
})), "HAC output 3 has incorrect data");
}
