LightGBM::DenseBin< VAL_T > Class Template Reference

Used to store bins for dense feature Use template to reduce memory cost. More...

#include <dense_bin.hpp>

Inheritance diagram for LightGBM::DenseBin< VAL_T >:

Public Member Functions
	DenseBin (data_size_t num_data)
void	Push (int, data_size_t idx, uint32_t value) override
	Push one record \pram tid Thread id.
void	ReSize (data_size_t num_data) override
BinIterator *	GetIterator (uint32_t min_bin, uint32_t max_bin, uint32_t default_bin) const override
	Get bin iterator of this bin for specific feature.
void	ConstructHistogram (const data_size_t *data_indices, data_size_t num_data, const score_t *ordered_gradients, const score_t *ordered_hessians, HistogramBinEntry *out) const override
	Construct histogram of this feature, Note: We use ordered_gradients and ordered_hessians to improve cache hit chance The naive solution is using gradients[data_indices[i]] for data_indices[i] to get gradients, which is not cache friendly, since the access of memory is not continuous. ordered_gradients and ordered_hessians are preprocessed, and they are re-ordered by data_indices. Ordered_gradients[i] is aligned with data_indices[i]'s gradients (same for ordered_hessians).
void	ConstructHistogram (data_size_t num_data, const score_t *ordered_gradients, const score_t *ordered_hessians, HistogramBinEntry *out) const override
void	ConstructHistogram (const data_size_t *data_indices, data_size_t num_data, const score_t *ordered_gradients, HistogramBinEntry *out) const override
	Construct histogram of this feature, Note: We use ordered_gradients and ordered_hessians to improve cache hit chance The naive solution is using gradients[data_indices[i]] for data_indices[i] to get gradients, which is not cache friendly, since the access of memory is not continuous. ordered_gradients and ordered_hessians are preprocessed, and they are re-ordered by data_indices. Ordered_gradients[i] is aligned with data_indices[i]'s gradients (same for ordered_hessians).
void	ConstructHistogram (data_size_t num_data, const score_t *ordered_gradients, HistogramBinEntry *out) const override
virtual data_size_t	Split (uint32_t min_bin, uint32_t max_bin, uint32_t default_bin, MissingType missing_type, bool default_left, uint32_t threshold, data_size_t *data_indices, data_size_t num_data, data_size_t *lte_indices, data_size_t *gt_indices) const override
	Split data according to threshold, if bin <= threshold, will put into left(lte_indices), else put into right(gt_indices)
virtual data_size_t	SplitCategorical (uint32_t min_bin, uint32_t max_bin, uint32_t default_bin, const uint32_t *threshold, int num_threahold, data_size_t *data_indices, data_size_t num_data, data_size_t *lte_indices, data_size_t *gt_indices) const override
	Split data according to threshold, if bin <= threshold, will put into left(lte_indices), else put into right(gt_indices)
data_size_t	num_data () const override
	Number of all data.
OrderedBin *	CreateOrderedBin () const override
	not ordered bin for dense feature
void	FinishLoad () override
	After pushed all feature data, call this could have better refactor for bin data.
void	LoadFromMemory (const void *memory, const std::vector< data_size_t > &local_used_indices) override
	Load from memory.
void	CopySubset (const Bin *full_bin, const data_size_t *used_indices, data_size_t num_used_indices) override
void	SaveBinaryToFile (const VirtualFileWriter *writer) const override
	Save binary data to file.
size_t	SizesInByte () const override
	Get sizes in byte of this object.
Public Member Functions inherited from LightGBM::Bin
virtual	~Bin ()
	virtual destructor

Data Fields
friend	DenseBinIterator< VAL_T >

Protected Attributes
data_size_t	num_data_
std::vector< VAL_T >	data_

Additional Inherited Members
Static Public Member Functions inherited from LightGBM::Bin
static Bin *	CreateBin (data_size_t num_data, int num_bin, double sparse_rate, bool is_enable_sparse, double sparse_threshold, bool *is_sparse)
	Create object for bin data of one feature, will call CreateDenseBin or CreateSparseBin according to "is_sparse".
static Bin *	CreateDenseBin (data_size_t num_data, int num_bin)
	Create object for bin data of one feature, used for dense feature.
static Bin *	CreateSparseBin (data_size_t num_data, int num_bin)
	Create object for bin data of one feature, used for sparse feature.

Detailed Description

template<typename VAL_T>
class LightGBM::DenseBin< VAL_T >

Used to store bins for dense feature Use template to reduce memory cost.

Member Function Documentation

ConstructHistogram() [1/4]

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::ConstructHistogram ( const data_size_t *  data_indices, data_size_t  num_data, const score_t *  ordered_gradients, const score_t *  ordered_hessians, HistogramBinEntry *  out  ) const

inlineoverridevirtual

Construct histogram of this feature, Note: We use ordered_gradients and ordered_hessians to improve cache hit chance The naive solution is using gradients[data_indices[i]] for data_indices[i] to get gradients, which is not cache friendly, since the access of memory is not continuous. ordered_gradients and ordered_hessians are preprocessed, and they are re-ordered by data_indices. Ordered_gradients[i] is aligned with data_indices[i]'s gradients (same for ordered_hessians).

Parameters

data_indices	Used data indices in current leaf
num_data	Number of used data
ordered_gradients	Pointer to gradients, the data_indices[i]-th data's gradient is ordered_gradients[i]
ordered_hessians	Pointer to hessians, the data_indices[i]-th data's hessian is ordered_hessians[i]
out	Output Result

Implements LightGBM::Bin.

ConstructHistogram() [2/4]

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::ConstructHistogram ( const data_size_t *  data_indices, data_size_t  num_data, const score_t *  ordered_gradients, HistogramBinEntry *  out  ) const

inlineoverridevirtual

Construct histogram of this feature, Note: We use ordered_gradients and ordered_hessians to improve cache hit chance The naive solution is using gradients[data_indices[i]] for data_indices[i] to get gradients, which is not cache friendly, since the access of memory is not continuous. ordered_gradients and ordered_hessians are preprocessed, and they are re-ordered by data_indices. Ordered_gradients[i] is aligned with data_indices[i]'s gradients (same for ordered_hessians).

Parameters

data_indices	Used data indices in current leaf
num_data	Number of used data
ordered_gradients	Pointer to gradients, the data_indices[i]-th data's gradient is ordered_gradients[i]
out	Output Result

Implements LightGBM::Bin.

ConstructHistogram() [3/4]

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::ConstructHistogram ( data_size_t  num_data, const score_t *  ordered_gradients, const score_t *  ordered_hessians, HistogramBinEntry *  out  ) const

inlineoverridevirtual

Implements LightGBM::Bin.

ConstructHistogram() [4/4]

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::ConstructHistogram ( data_size_t  num_data, const score_t *  ordered_gradients, HistogramBinEntry *  out  ) const

inlineoverridevirtual

Implements LightGBM::Bin.

CopySubset()

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::CopySubset ( const Bin *  full_bin, const data_size_t *  used_indices, data_size_t  num_used_indices  )

inlineoverridevirtual

Implements LightGBM::Bin.

CreateOrderedBin()

template<typename VAL_T >

OrderedBin * LightGBM::DenseBin< VAL_T >::CreateOrderedBin ( ) const

inlineoverridevirtual

not ordered bin for dense feature

Implements LightGBM::Bin.

FinishLoad()

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::FinishLoad ( )

inlineoverridevirtual

After pushed all feature data, call this could have better refactor for bin data.

Implements LightGBM::Bin.

GetIterator()

template<typename VAL_T >

BinIterator * LightGBM::DenseBin< VAL_T >::GetIterator ( uint32_t  min_bin, uint32_t  max_bin, uint32_t  default_bin  ) const

overridevirtual

Get bin iterator of this bin for specific feature.

Parameters

min_bin	min_bin of current used feature
max_bin	max_bin of current used feature
default_bin	default bin if bin not in [min_bin, max_bin]

Returns

Iterator of this bin

Implements LightGBM::Bin.

LoadFromMemory()

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::LoadFromMemory ( const void *  memory, const std::vector< data_size_t > &  local_used_indices  )

inlineoverridevirtual

Load from memory.

Parameters

memory
local_used_indices

Implements LightGBM::Bin.

num_data()

template<typename VAL_T >

data_size_t LightGBM::DenseBin< VAL_T >::num_data ( ) const

inlineoverridevirtual

Number of all data.

Implements LightGBM::Bin.

Push()

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::Push ( int  tid, data_size_t  idx, uint32_t  value  )

inlineoverridevirtual

Push one record \pram tid Thread id.

Parameters

idx	Index of record
value	bin value of record

Implements LightGBM::Bin.

ReSize()

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::ReSize ( data_size_t  num_data )

inlineoverridevirtual

Implements LightGBM::Bin.

SaveBinaryToFile()

template<typename VAL_T >

void LightGBM::DenseBin< VAL_T >::SaveBinaryToFile ( const VirtualFileWriter *  writer ) const

inlineoverridevirtual

Save binary data to file.

Parameters

file	File want to write

Implements LightGBM::Bin.

SizesInByte()

template<typename VAL_T >

size_t LightGBM::DenseBin< VAL_T >::SizesInByte ( ) const

inlineoverridevirtual

Get sizes in byte of this object.

Implements LightGBM::Bin.

Split()

template<typename VAL_T >

virtual data_size_t LightGBM::DenseBin< VAL_T >::Split ( uint32_t  min_bin, uint32_t  max_bin, uint32_t  default_bin, MissingType  missing_type, bool  default_left, uint32_t  threshold, data_size_t *  data_indices, data_size_t  num_data, data_size_t *  lte_indices, data_size_t *  gt_indices  ) const

inlineoverridevirtual

Split data according to threshold, if bin <= threshold, will put into left(lte_indices), else put into right(gt_indices)

Parameters

min_bin	min_bin of current used feature
max_bin	max_bin of current used feature
default_bin	defualt bin if bin not in [min_bin, max_bin]
missing_type	missing type
default_left	missing bin will go to left child
threshold	The split threshold.
data_indices	Used data indices. After called this function. The less than or equal data indices will store on this object.
num_data	Number of used data
lte_indices	After called this function. The less or equal data indices will store on this object.
gt_indices	After called this function. The greater data indices will store on this object.

Returns

The number of less than or equal data.

Implements LightGBM::Bin.

SplitCategorical()

template<typename VAL_T >

virtual data_size_t LightGBM::DenseBin< VAL_T >::SplitCategorical ( uint32_t  min_bin, uint32_t  max_bin, uint32_t  default_bin, const uint32_t *  threshold, int  num_threshold, data_size_t *  data_indices, data_size_t  num_data, data_size_t *  lte_indices, data_size_t *  gt_indices  ) const

inlineoverridevirtual

Split data according to threshold, if bin <= threshold, will put into left(lte_indices), else put into right(gt_indices)

Parameters

min_bin	min_bin of current used feature
max_bin	max_bin of current used feature
default_bin	defualt bin if bin not in [min_bin, max_bin]
threshold	The split threshold.
num_threshold	Number of threshold
data_indices	Used data indices. After called this function. The less than or equal data indices will store on this object.
num_data	Number of used data
lte_indices	After called this function. The less or equal data indices will store on this object.
gt_indices	After called this function. The greater data indices will store on this object.

Returns

The number of less than or equal data.

Implements LightGBM::Bin.

---

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

• External/LightGBM_2.2.3/LightGBM-2.2.3/src/io/dense_bin.hpp