public Histogram(BucketSize bucketSize, IEnumerable <HistogramBucket> buckets)
{
Debug.Assert(
buckets.All(b => b.BucketSize.SnappedSize == bucketSize.SnappedSize),
$"Histogram buckets don't match given bucketsize");
BucketSize = bucketSize;
Buckets = new SortedList <decimal, HistogramBucket>(buckets.ToDictionary(b => b.LowerBound));
}
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[]
{
bucketSizeEstimate.Smaller(steps: 2),
bucketSizeEstimate,
bucketSizeEstimate.Larger(steps: 2),
}
.Where(b => !(b is null))
.Select(b => b !.SnappedSize)
.Distinct()
.ToArray());
}
internal HistogramBucket(decimal lowerBound, BucketSize bucketSize, NoisyCount noisyCount)
{
LowerBound = lowerBound;
BucketSize = bucketSize;
this.noisyCount = noisyCount;
}
public bool Equals(HistogramBucket bucket) => noisyCount.Equals(bucket.noisyCount) && BucketSize.Equals(bucket.BucketSize) && LowerBound == bucket.LowerBound && Count == bucket.Count && CountNoise == bucket.CountNoise;
