turi::group_aggregate_value Class Reference

#include <core/storage/sframe_data/group_aggregate_value.hpp>

Public Member Functions
virtual group_aggregate_value *	new_instance () const =0
virtual void	add_element (const std::vector< flexible_type > &values)
virtual void	add_element_simple (const flexible_type &flex)=0
virtual void	partial_finalize ()
virtual void	combine (const group_aggregate_value &other)=0
virtual flexible_type	emit () const =0
virtual bool	support_type (flex_type_enum type) const =0
virtual flex_type_enum	set_input_types (const std::vector< flex_type_enum > &types)
virtual std::string	name () const =0
virtual void	save (oarchive &oarc) const =0
virtual void	load (iarchive &iarc)=0
virtual	~group_aggregate_value ()
virtual void	print (std::ostream &os) const

Detailed Description

The base class specification for representing the intermediate state as well as the computation (aggregation, combining and output) for an aggregation operation.

Essentially, the group_aggregate_value must describe a parallel reduction in the following form:

for streamid in data_stream:
  for data in stream[streamid]:
    value[streamid].add_element(data)
  value[streamid].partial_finalize()

for streamid in data_stream:
  final_value.combine(value[streamid])

output = final_value.emit()

Each value could have an arbitrary number of elements inserted into it. When all insertions are complete, partial_finalize() is called on the value. Values can be combined with each other to form a final value, which then emits a response.

Definition at line 45 of file group_aggregate_value.hpp.

Constructor & Destructor Documentation

~group_aggregate_value()

virtual turi::group_aggregate_value::~group_aggregate_value ( )

inlinevirtual

Destruction

Definition at line 125 of file group_aggregate_value.hpp.

Member Function Documentation

add_element()

virtual void turi::group_aggregate_value::add_element ( const std::vector< flexible_type > &  values )

inlinevirtual

Adds an element to the aggregate. Elements to be added will be either the input_type (as set by set_input_type()) or UNDEFINED.

Operator that expects more than one input values need to overwrite this function

Reimplemented in turi::groupby_operators::zip_dict, turi::groupby_operators::argmax, and turi::groupby_operators::argmin.

Definition at line 60 of file group_aggregate_value.hpp.

add_element_simple()

virtual void turi::group_aggregate_value::add_element_simple ( const flexible_type &  flex )

pure virtual

Adds an element to the aggregate. Simple version of add_element where there is only one input value for the operator

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::query_eval::generic_aggregator< T, AggregateFunctionType >, and turi::groupby_operators::vector_sum.

combine()

virtual void turi::group_aggregate_value::combine ( const group_aggregate_value &  other )

pure virtual

Combines two partial aggregates

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::groupby_operators::vector_sum, and turi::query_eval::generic_aggregator< T, AggregateFunctionType >.

emit()

virtual flexible_type turi::group_aggregate_value::emit ( ) const

pure virtual

Emits the result of a groupby operation

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::distinct, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::groupby_operators::vector_sum, and turi::query_eval::generic_aggregator< T, AggregateFunctionType >.

load()

virtual void turi::group_aggregate_value::load ( iarchive &  iarc )

pure virtual

Loads the state of the aggregation from an archive

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::groupby_operators::vector_sum, and turi::query_eval::generic_aggregator< T, AggregateFunctionType >.

name()

virtual std::string turi::group_aggregate_value::name ( ) const

pure virtual

Returns a printable name of the operation.

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::distinct, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::groupby_operators::vector_sum, and turi::query_eval::generic_aggregator< T, AggregateFunctionType >.

new_instance()

virtual group_aggregate_value* turi::group_aggregate_value::new_instance ( ) const

pure virtual

Creates a new instance of the aggregator. The new instance must remember the input type (see set_input_type()) but have have a new empty value.

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::distinct, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::query_eval::generic_aggregator< T, AggregateFunctionType >, and turi::groupby_operators::vector_sum.

partial_finalize()

virtual void turi::group_aggregate_value::partial_finalize ( )

inlinevirtual

No more elements will be added to this value. However, this value may still be combined with other values.

Reimplemented in turi::groupby_operators::quantile.

Definition at line 75 of file group_aggregate_value.hpp.

print()

virtual void turi::group_aggregate_value::print ( std::ostream &  os ) const

inlinevirtual

Override this function for allowing the operator to be easily printed.

std::cout << aggregator <<s std::endl;

Reimplemented in turi::groupby_operators::variance.

Definition at line 134 of file group_aggregate_value.hpp.

save()

virtual void turi::group_aggregate_value::save ( oarchive &  oarc ) const

pure virtual

Saves the state of the aggregation to an archive

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::groupby_operators::vector_sum, and turi::query_eval::generic_aggregator< T, AggregateFunctionType >.

set_input_types()

virtual flex_type_enum turi::group_aggregate_value::set_input_types ( const std::vector< flex_type_enum > &  types )

inlinevirtual

Sets the input types and returns the output type. For instance, a sum aggregator when summing integers will return an integer, and when summing doubles will return doubles.

Default implementation assumes there is ony one input, and output type is the same as input type.

Reimplemented in turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::argmax, and turi::groupby_operators::argmin.

Definition at line 102 of file group_aggregate_value.hpp.

support_type()

virtual bool turi::group_aggregate_value::support_type ( flex_type_enum  type ) const

pure virtual

Returns true if the the aggregate_value can consume a column of this type, and false otherwise. (For instance, a sum aggregator can consume integers and floats, and not anything else).

Implemented in turi::groupby_operators::frequency_count, turi::groupby_operators::count_distinct, turi::groupby_operators::select_one, turi::groupby_operators::zip_list, turi::groupby_operators::zip_dict, turi::groupby_operators::quantile, turi::groupby_operators::variance, turi::groupby_operators::average, turi::groupby_operators::vector_average, turi::groupby_operators::non_null_count, turi::groupby_operators::count, turi::groupby_operators::max, turi::groupby_operators::argmax, turi::groupby_operators::argmin, turi::groupby_operators::min, turi::groupby_operators::sum, turi::groupby_operators::vector_sum, and turi::query_eval::generic_aggregator< T, AggregateFunctionType >.

---

The documentation for this class was generated from the following file:

• core/storage/sframe_data/group_aggregate_value.hpp