public static CountMinSketchState MakeState(CountMinSketchSpec spec)
{
CountMinSketchStateHashes hashes = CountMinSketchStateHashes.MakeState(spec.HashesSpec);
CountMinSketchStateTopk topk = null;
if (spec.TopkSpec != null && spec.TopkSpec > 0)
{
topk = new CountMinSketchStateTopk(spec.TopkSpec.Value);
}
return(new CountMinSketchState(hashes, topk));
}
public CountMinSketchState(CountMinSketchStateHashes hashes, CountMinSketchStateTopk topk)
{
Hashes = hashes;
Topk = topk;
}
public void TestTopK()
{
const int space = 10000;
const int points = 100000;
const int topkMax = 100;
var random = new Random();
var topk = new CountMinSketchStateTopk(topkMax);
var sent = new Dictionary <Blob, long?>();
for (var i = 0; i < points; i++)
{
// for simple population: Blob bytes = generateBytesModulo(i, space);
var bytes = GenerateBytesRandom(random, space);
var count = sent.Get(bytes);
if (count == null)
{
sent.Put(bytes, 1L);
topk.UpdateExpectIncreasing(bytes.Data, 1);
}
else
{
sent.Put(bytes, count + 1);
topk.UpdateExpectIncreasing(bytes.Data, count.Value + 1);
}
if (i > 0 && i % 100000 == 0)
{
Console.WriteLine("Completed {0}", i);
}
}
// compare
var top = topk.TopKValues;
// assert filled
if (sent.Count < topkMax)
{
Assert.AreEqual(sent.Count, top.Count);
}
else
{
Assert.AreEqual(topkMax, top.Count);
}
// assert no duplicate values
var set = new HashSet <Blob>();
foreach (var topBytes in top)
{
Assert.IsTrue(set.Add(topBytes));
}
// assert order descending
long?lastFreq = null;
foreach (var topBytes in top)
{
var freq = sent.Get(topBytes);
if (lastFreq != null)
{
Assert.IsTrue(freq <= lastFreq);
}
lastFreq = freq;
}
}