This class implements the main logic of the quickrank application. More...

#include <driver.h>

Public Member Functions
	Driver ()

virtual	~Driver ()

Static Public Member Functions
static int	run (ParamsMap &pmap)
	Implements the main logic of the quickrank application, detecting the metrics to adopt and the phases to execute (train/validation/test). More...

Static Private Member Functions
static void	training_phase (std::shared_ptr< learning::LTR_Algorithm > algo, std::shared_ptr< metric::ir::Metric > train_metric, std::shared_ptr< quickrank::data::Dataset > training_dataset, std::shared_ptr< quickrank::data::Dataset > validation_dataset, const std::string output_filename, const size_t npartialsave)
	Runs train/validation of algo by optimizing train_metric and then measures test_metric on the test data. More...

static void	optimization_phase (std::shared_ptr< quickrank::optimization::Optimization > opt_algorithm, std::shared_ptr< learning::LTR_Algorithm > ranking_algo, std::shared_ptr< metric::ir::Metric > train_metric, std::shared_ptr< quickrank::data::Dataset > training_dataset, std::shared_ptr< quickrank::data::Dataset > validation_dataset, std::string training_partial_filename, std::string validation_partial_filename, const std::string output_filename, const std::string opt_algo_model_filename, const size_t npartialsave)
	Runs train/validation of algo by optimizing train_metric and then measures test_metric on the test data. More...

static void	testing_phase (std::shared_ptr< learning::LTR_Algorithm > algo, std::shared_ptr< metric::ir::Metric > test_metric, std::shared_ptr< quickrank::data::Dataset > test_dataset, const std::string scores_filename, const bool detailed_testing)
	Runs the learned or loaded model on the test data and then measures test_metric on the test data. More...

static std::shared_ptr< quickrank::data::Dataset >	load_dataset (const std::string dataset_filename, const std::string dataset_label)

static std::shared_ptr< data::Dataset >	extract_partial_scores (std::shared_ptr< learning::LTR_Algorithm > algo, std::shared_ptr< data::Dataset > input_dataset)

Detailed Description

This class implements the main logic of the quickrank application.

Constructor & Destructor Documentation

quickrank::driver::Driver::Driver ( )

quickrank::driver::Driver::~Driver ( )

virtual

Member Function Documentation

std::shared_ptr< data::Dataset > quickrank::driver::Driver::extract_partial_scores	(	std::shared_ptr< learning::LTR_Algorithm >	algo,
		std::shared_ptr< data::Dataset >	input_dataset
	)

staticprivate

std::shared_ptr< quickrank::data::Dataset > quickrank::driver::Driver::load_dataset	(	const std::string	dataset_filename,
		const std::string	dataset_label
	)

staticprivate

void quickrank::driver::Driver::optimization_phase	(	std::shared_ptr< quickrank::optimization::Optimization >	opt_algorithm,
		std::shared_ptr< learning::LTR_Algorithm >	ranking_algo,
		std::shared_ptr< metric::ir::Metric >	train_metric,
		std::shared_ptr< quickrank::data::Dataset >	training_dataset,
		std::shared_ptr< quickrank::data::Dataset >	validation_dataset,
		std::string	training_partial_filename,
		std::string	validation_partial_filename,
		const std::string	output_filename,
		const std::string	opt_algo_model_filename,
		const size_t	npartialsave
	)

staticprivate

Runs train/validation of algo by optimizing train_metric and then measures test_metric on the test data.

Parameters

algo	The L-T-R algorithm to be tested.
train_metric	The metric optimized during training.
training_filename	The training dataset.
validation_filename	The validation dataset. If empty, validation is not used.
output_filename	Model output file. If empty, no output file is written.
npartialsave	Allows to save a partial model every given number of iterations.

int quickrank::driver::Driver::run ( ParamsMap & pmap )

static

Implements the main logic of the quickrank application, detecting the metrics to adopt and the phases to execute (train/validation/test).

Returns the exit code of the application

Parameters

void quickrank::driver::Driver::testing_phase	(	std::shared_ptr< learning::LTR_Algorithm >	algo,
		std::shared_ptr< metric::ir::Metric >	test_metric,
		std::shared_ptr< quickrank::data::Dataset >	test_dataset,
		const std::string	scores_filename,
		const bool	detailed_testing
	)

staticprivate

Runs the learned or loaded model on the test data and then measures test_metric on the test data.

Parameters

algo	The L-T-R algorithm to be tested.
test_metric	The metric measured on the test data.
test_filename	The test dataset. If empty, no performance is measured on the test set.
scores_filename	The output scores file. If set save the scores computed for the test set.
verbose	If True saves an SVML-like file with the score of each ranker in the ensemble. NB. Works only for ensembles.

void quickrank::driver::Driver::training_phase	(	std::shared_ptr< learning::LTR_Algorithm >	algo,
		std::shared_ptr< metric::ir::Metric >	train_metric,
		std::shared_ptr< quickrank::data::Dataset >	training_dataset,
		std::shared_ptr< quickrank::data::Dataset >	validation_dataset,
		const std::string	output_filename,
		const size_t	npartialsave
	)

staticprivate

Runs train/validation of algo by optimizing train_metric and then measures test_metric on the test data.

Parameters

algo	The L-T-R algorithm to be tested.
train_metric	The metric optimized during training.
training_filename	The training dataset.
validation_filename	The validation dataset. If empty, validation is not used.
output_filename	Model output file. If empty, no output file is written.
npartialsave	Allows to save a partial model every given number of iterations.

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

include/driver/driver.h
src/driver/driver.cc

Public Member Functions

Static Public Member Functions

Static Private Member Functions

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation