2012-10-29 02:58:52 +00:00
|
|
|
|
#pragma once
|
|
|
|
|
|
|
|
|
|
#include <stats/ReservoirSampler.h>
|
|
|
|
|
|
|
|
|
|
#include <DB/IO/WriteHelpers.h>
|
|
|
|
|
#include <DB/IO/ReadHelpers.h>
|
|
|
|
|
|
|
|
|
|
#include <DB/DataTypes/DataTypesNumberFixed.h>
|
|
|
|
|
|
|
|
|
|
#include <DB/AggregateFunctions/IUnaryAggregateFunction.h>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
namespace DB
|
|
|
|
|
{
|
|
|
|
|
|
2013-02-08 19:34:44 +00:00
|
|
|
|
template <typename ArgumentFieldType, bool returns_float = true>
|
|
|
|
|
struct AggregateFunctionQuantileData
|
|
|
|
|
{
|
|
|
|
|
typedef ReservoirSampler<ArgumentFieldType> Sample;
|
|
|
|
|
Sample sample;
|
|
|
|
|
};
|
|
|
|
|
|
2012-10-29 02:58:52 +00:00
|
|
|
|
|
|
|
|
|
/** Приближённо вычисляет квантиль.
|
|
|
|
|
* В качестве типа аргумента может быть только числовой тип (в том числе, дата и дата-с-временем).
|
|
|
|
|
* Если returns_float = true, то типом результата будет Float64, иначе - тип результата совпадает с типом аргумента.
|
|
|
|
|
* Для дат и дат-с-временем returns_float следует задавать равным false.
|
|
|
|
|
*/
|
|
|
|
|
template <typename ArgumentFieldType, bool returns_float = true>
|
2013-02-08 19:34:44 +00:00
|
|
|
|
class AggregateFunctionQuantile : public IUnaryAggregateFunction<AggregateFunctionQuantileData<ArgumentFieldType, returns_float> >
|
2012-10-29 02:58:52 +00:00
|
|
|
|
{
|
|
|
|
|
private:
|
|
|
|
|
typedef AggregateFunctionQuantile<ArgumentFieldType, returns_float> Self;
|
|
|
|
|
typedef ReservoirSampler<ArgumentFieldType> Sample;
|
|
|
|
|
|
|
|
|
|
double level;
|
|
|
|
|
DataTypePtr type;
|
|
|
|
|
|
|
|
|
|
public:
|
|
|
|
|
AggregateFunctionQuantile(double level_ = 0.5) : level(level_) {}
|
|
|
|
|
|
|
|
|
|
String getName() const { return "quantile"; }
|
|
|
|
|
String getTypeID() const { return (returns_float ? "quantile_float_" : "quantile_rounded_") + TypeName<ArgumentFieldType>::get(); }
|
|
|
|
|
|
|
|
|
|
DataTypePtr getReturnType() const
|
|
|
|
|
{
|
|
|
|
|
return type;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void setArgument(const DataTypePtr & argument)
|
|
|
|
|
{
|
|
|
|
|
if (returns_float)
|
|
|
|
|
type = new DataTypeFloat64;
|
|
|
|
|
else
|
|
|
|
|
type = argument;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void setParameters(const Row & params)
|
|
|
|
|
{
|
|
|
|
|
if (params.size() != 1)
|
|
|
|
|
throw Exception("Aggregate function " + getName() + " requires exactly one parameter.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
|
|
|
|
|
|
2013-01-05 20:03:19 +00:00
|
|
|
|
level = apply_visitor(FieldVisitorConvertToNumber<Float64>(), params[0]);
|
2012-10-29 02:58:52 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-02-08 19:34:44 +00:00
|
|
|
|
|
|
|
|
|
void addOne(AggregateDataPtr place, const Field & value_) const
|
2012-10-29 02:58:52 +00:00
|
|
|
|
{
|
2013-02-08 19:34:44 +00:00
|
|
|
|
data(place).sample.insert(get<typename NearestFieldType<ArgumentFieldType>::Type>(value));
|
2012-10-29 02:58:52 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-02-08 19:34:44 +00:00
|
|
|
|
void merge(AggregateDataPtr place, ConstAggregateDataPtr rhs) const
|
2012-10-29 02:58:52 +00:00
|
|
|
|
{
|
2013-02-08 19:34:44 +00:00
|
|
|
|
data(place).sample.merge(data(rhs).sample);
|
2012-10-29 02:58:52 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-02-08 19:34:44 +00:00
|
|
|
|
void serialize(ConstAggregateDataPtr place, WriteBuffer & buf) const
|
2012-10-29 02:58:52 +00:00
|
|
|
|
{
|
2013-02-08 19:34:44 +00:00
|
|
|
|
data(place).sample.write(buf);
|
2012-10-29 02:58:52 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-02-08 19:34:44 +00:00
|
|
|
|
void deserializeMerge(AggregateDataPtr place, ReadBuffer & buf) const
|
2012-10-29 02:58:52 +00:00
|
|
|
|
{
|
|
|
|
|
Sample tmp_sample;
|
|
|
|
|
tmp_sample.read(buf);
|
2013-02-08 19:34:44 +00:00
|
|
|
|
data(place).sample.merge(tmp_sample);
|
2012-10-29 02:58:52 +00:00
|
|
|
|
}
|
|
|
|
|
|
2013-02-08 19:34:44 +00:00
|
|
|
|
Field getResult(ConstAggregateDataPtr place) const
|
2012-10-29 02:58:52 +00:00
|
|
|
|
{
|
2012-11-14 09:38:30 +00:00
|
|
|
|
/// Sample может отсортироваться при получении квантиля, но в этом контексте можно не считать это нарушением константности.
|
2012-10-29 02:58:52 +00:00
|
|
|
|
|
|
|
|
|
if (returns_float)
|
2013-02-08 19:34:44 +00:00
|
|
|
|
return Float64(const_cast<Sample &>(data(place).sample).quantileInterpolated(level));
|
2012-10-29 02:58:52 +00:00
|
|
|
|
else
|
2013-02-08 19:34:44 +00:00
|
|
|
|
return typename NearestFieldType<ArgumentFieldType>::Type(const_cast<Sample &>(data(place).sample).quantileInterpolated(level));
|
2012-10-29 02:58:52 +00:00
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
}
|