internal static decimal[] EstimateBucketResolutions(
long numSamples,
double minSample,
double maxSample,
long valuesPerBucketTarget,
bool isIntegerColumn)
{
if (numSamples <= 0)
{
throw new System.ArgumentException(
$"Argument numSamples should always be greater than zero, got {numSamples}.");
}
var range = maxSample - minSample;
if (range <= 0)
{
return(new decimal[] { 1M });
}
var valueDensity = numSamples / (maxSample - minSample);
var targetBucketSize = valuesPerBucketTarget / valueDensity;
var bucketSizeEstimate = new BucketSize(isIntegerColumn
? System.Math.Max(targetBucketSize, 5)
: targetBucketSize);
return(new decimal[]
{
bucketSizeEstimate.Smaller(steps: 2).SnappedSize,
bucketSizeEstimate.SnappedSize,
bucketSizeEstimate.Larger(steps: 2).SnappedSize,
});
}
public void TestSmaller()
{
var b = new BucketSize(80);
Assert.Equal(100M, b.SnappedSize);
Assert.Equal(20M, b.Smaller(2).SnappedSize);
}
public void TestSmallerOutOfBounds()
{
var b = new BucketSize(0.00022M);
Assert.Equal(0.0002M, b.SnappedSize);
Assert.Equal(0.0001M, b.Smaller(2).SnappedSize);
}
public TlshBuilder(BucketSize bucketSize, ChecksumSize checksumSize)
{
bucketCount = (int)bucketSize;
checksumLength = (int)checksumSize;
// Each bucket => 2 bits of output code.
codeSize = bucketCount >> 2;
slideWindow = new int[CSlidingWindowSize];
accumulatorBuckets = new uint[CBuckets];
if (checksumLength > 1)
{
checksumArray = new int[checksumLength];
}
}
public static DateTime DecreaseDateTimeBy(DateTime toAdd, BucketSize bucketSize)
{
if (bucketSize == BucketSize.Day)
{
return(toAdd.AddDays(-1));
}
if (bucketSize == BucketSize.Month)
{
return(toAdd.AddMonths(-1));
}
if (bucketSize == BucketSize.Year)
{
return(toAdd.AddYears(-1));
}
throw new NotSupportedException("Unknown bucket size " + bucketSize);
}
public bool IsTimeInBucket(DateTime toCheck, BucketSize bucketSize)
{
DateTime upperLimit = Time;
if (bucketSize == BucketSize.Day)
{
upperLimit = upperLimit.AddDays(1);
}
else
if (bucketSize == BucketSize.Month)
{
upperLimit = upperLimit.AddMonths(1);
}
else
if (bucketSize == BucketSize.Year)
{
upperLimit = upperLimit.AddYears(1);
}
return(toCheck >= Time && toCheck < upperLimit);
}
internal HistogramBucket(decimal lowerBound, BucketSize bucketSize, NoisyCount noisyCount)
{
LowerBound = lowerBound;
BucketSize = bucketSize;
this.noisyCount = noisyCount;
}
public void TestSnap200()
{
var b = new BucketSize(150);
Assert.Equal(200M, b.SnappedSize);
}
public void TestSnap100()
{
var b = new BucketSize(120M);
Assert.Equal(100M, b.SnappedSize);
}
public void TestSnapSecond()
{
var b = new BucketSize(0.00016M);
Assert.Equal(0.0002M, b.SnappedSize);
}
public void TestSnapFirst()
{
var b = new BucketSize(0.00014M);
Assert.Equal(0.0001M, b.SnappedSize);
}
public void TestSmallValue()
{
var b = new BucketSize(0.000001M);
Assert.Equal(0.0001M, b.SnappedSize);
}
public static DateTime RoundDateTimeDownToNearestBucketFloor(DateTime toRound, BucketSize bucketSize)
{
if (bucketSize == BucketSize.Day)
{
return(new DateTime(toRound.Year, toRound.Month, toRound.Day));
}
if (bucketSize == BucketSize.Month)
{
return(new DateTime(toRound.Year, toRound.Month, 1));
}
if (bucketSize == BucketSize.Year)
{
return(new DateTime(toRound.Year, 1, 1));
}
throw new NotSupportedException("Unknown bucket size " + bucketSize);
}
private static int HashStringLength(BucketSize bucketSize, ChecksumSize checksumSize)
{
return(((int)bucketSize / 2) + ((int)checksumSize * 2) + 4);
}
