private static Bucket sumBucket(Collection<Bucket> dist, double numGroups) {
Constant sumLow = null, sumHigh = new DoubleConstant(1.0);
double totalFreq = 0.0;
Map<Constant, Bucket> highs = new HashMap<Constant, Bucket>();
for (Bucket bkt : dist) {
// estimate sumLow as the only one smallest value in a group
if (sumLow == null || bkt.valueRange().low().compareTo(sumLow) < 0)
sumLow = bkt.valueRange().low();
totalFreq += bkt.frequency();
highs.put(bkt.valueRange().high(), bkt);
}
SortedSet<Constant> desc = new TreeSet<Constant>(highs.keySet()).descendingSet();
// estimate sumHigh as the sum of top maxGroupSize values
double maxGroupSize = totalFreq - numGroups + 1;
double currSize = 0.0;
for (Constant high : desc) {
Bucket bkt = highs.get(high);
double recsToSum = Math.min(bkt.frequency(), maxGroupSize - currSize);
sumHigh = sumHigh.add(high.mul(new DoubleConstant(recsToSum)));
currSize += recsToSum;
if (Double.compare(currSize, maxGroupSize) >= 0) break;
}
ConstantRange sumRange = ConstantRange.newInstance(sumLow, true, sumHigh, true);
// discard percentiles
return new Bucket(sumRange, numGroups, numGroups);
}
private static Bucket countBucket(Collection<Bucket> dist, double numGroups) {
Constant cntLow = new DoubleConstant(1.0);
Double totalFreq = 0.0;
for (Bucket bkt : dist) {
totalFreq += bkt.frequency();
}
double maxGroupSize = totalFreq - numGroups + 1;
Constant cntHigh = new DoubleConstant(maxGroupSize);
ConstantRange countRange = ConstantRange.newInstance(cntLow, true, cntHigh, true);
// discard percentiles
return new Bucket(countRange, numGroups, numGroups);
}
private static Bucket avgBucket(Collection<Bucket> dist, double numGroups) {
Constant avgLow = null, avgHigh = null;
for (Bucket bkt : dist) {
if (avgLow == null || bkt.valueRange().low().compareTo(avgLow) < 0)
avgLow = bkt.valueRange().low();
if (avgHigh == null || bkt.valueRange().high().compareTo(avgHigh) > 0)
avgHigh = bkt.valueRange().high();
}
ConstantRange avgRange = ConstantRange.newInstance(avgLow, true, avgHigh, true);
// discard percentiles
return new Bucket(avgRange, numGroups, numGroups);
}
private static Bucket distinctCountBucket(Collection<Bucket> dist, double numGroups) {
Constant dcLow = new DoubleConstant(1.0);
Double totalFreq = 0.0, dv = 0.0;
for (Bucket bkt : dist) {
totalFreq += bkt.frequency();
dv += bkt.distinctValues();
}
double maxGroupSize = totalFreq - numGroups + 1;
Constant dcHigh = new DoubleConstant(Math.min(maxGroupSize, dv));
ConstantRange distinctCountRange = ConstantRange.newInstance(dcLow, true, dcHigh, true);
// discard percentiles
return new Bucket(distinctCountRange, numGroups, numGroups);
}
private static Bucket maxBucket(Collection<Bucket> dist, double numGroups) {
Constant maxLow = null, maxHigh = null;
Double dv = 0.0;
for (Bucket bkt : dist) {
if (maxLow == null || bkt.valueRange().low().compareTo(maxLow) < 0)
maxLow = bkt.valueRange().low();
if (maxHigh == null || bkt.valueRange().high().compareTo(maxHigh) > 0)
maxHigh = bkt.valueRange().high();
dv += bkt.distinctValues();
}
ConstantRange maxRange = ConstantRange.newInstance(maxLow, true, maxHigh, true);
// discard percentiles
return new Bucket(maxRange, numGroups, Math.min(numGroups, dv));
}
