CNum::Data Namespace Reference

Tools used for gathering and grouping datasets. More...

Classes
struct	Bin
	A bin for quantile and uniform binning. More...
struct	Shelf
	Contains bins and the ranges of values they represent. More...

Functions
std::array< CNum::DataStructs::Matrix< double >, 2 >	get_data (std::string data_path, char seperator=',')
	Get data from a _SV file with last column being the labels.
void	PCA (std::string input_path, std::string output_path)
	Principle component analysis.
std::shared_ptr< Shelf[]>	uniform_bin (const CNum::DataStructs::Matrix< double > &data, size_t num_bins=256)
	Uniform binning of data.
std::shared_ptr< Shelf[]>	quantile_bin (const CNum::DataStructs::Matrix< double > &data, size_t num_bins=256)
	Quantile sketch not exact quantile bins.
CNum::DataStructs::Matrix< int >	apply_quantile (const CNum::DataStructs::Matrix< double > &data, std::shared_ptr< Shelf[]> shelves)
	Construct data matrix of bin values.

Detailed Description

Tools used for gathering and grouping datasets.

Function Documentation

apply_quantile()

CNum::DataStructs::Matrix< int > CNum::Data::apply_quantile	(	const CNum::DataStructs::Matrix< double > &	data,
		std::shared_ptr< Shelf[]>	shelves )

Construct data matrix of bin values.

Parameters

data	The dataset
shelves	The bins and the boundaries associated with them

Returns

The matrix of bin values

get_data()

std::array< CNum::DataStructs::Matrix< double >, 2 > CNum::Data::get_data	(	std::string	data_path,
		char	seperator = ',' )

Get data from a _SV file with last column being the labels.

Parameters

data_path	The path to the data file
seperator	The delimiter of the columns

Returns

The data and labels [data, labels]

PCA()

void CNum::Data::PCA	(	std::string	input_path,
		std::string	output_path )

Principle component analysis.

Available next release

quantile_bin()

std::shared_ptr< Shelf[]> CNum::Data::quantile_bin	(	const CNum::DataStructs::Matrix< double > &	data,
		size_t	num_bins = 256 )

Quantile sketch not exact quantile bins.

As perfect quantile binning is an expensive operation I used a quantile binning approximation

Parameters

data	The dataset
num_bins	The number of bins to distribute the data among

Returns

The bins and the boundaries associated with them

uniform_bin()

std::shared_ptr< Shelf[]> CNum::Data::uniform_bin	(	const CNum::DataStructs::Matrix< double > &	data,
		size_t	num_bins = 256 )

Uniform binning of data.

Parameters

data	The dataset
num_bins	The number of bins to distribute the data among

Returns

The bins and the boundaries associated with them