ProfileInitializer Class Reference

Profile initializer heuristic. More...

#include <ProfileInitializer.h>

Public Member Functions
	ProfileInitializer (Config &config, AlternativesPerformance &altPerfs)
	ProfileInitializer (const ProfileInitializer &profInit)
	~ProfileInitializer ()
AlternativesPerformance	getAlternativesPerformance () const
void	setAlternativesPerformance (AlternativesPerformance &newAltPerfs)
std::vector< float >	categoryFrequency ()
std::vector< Perf >	getProfilePerformanceCandidates (const Criterion &crit, const Category &cat, const int nbCategories)
float	weightedProbability (const Perf perfAlt, const Criterion &crit, const Category &catAbove, const Category &catBelow, const int nbCategories, const std::vector< float > &catFrequency, std::vector< Perf > &candidates, float delta=0.001)
std::vector< Perf >	initializeProfilePerformance (const Criterion &crit, Categories &categories, const std::vector< float > &catFre)
void	initializeProfiles (MRSortModel &model)

Detailed Description

Profile initializer heuristic.

This class implements the profile initializer heuristic as described in the thesis. This algorithm is initialized with the dataset and can then be used to transform a model into an "initialized" one.

A complete description of the heuristic can be found in Profile Initialization.

Constructor & Destructor Documentation

ProfileInitializer() [1/2]

ProfileInitializer::ProfileInitializer	(	Config &	config,
		AlternativesPerformance &	altPerfs
	)

ProfileInitializer standard constructor.

Parameters

altPerfs	AlternativesPerformance objet that represents our dataset
config	config setup for logging

ProfileInitializer() [2/2]

ProfileInitializer::ProfileInitializer

(

const ProfileInitializer &

profInit

)

ProfileInitializer standard copy constructor.

Parameters

profInit	ProfileInitializer object
config	config setup for logging

~ProfileInitializer()

ProfileInitializer::~ProfileInitializer

(

)

ProfileInitializer standard deconstructor

Member Function Documentation

categoryFrequency()

std::vector<float> ProfileInitializer::categoryFrequency

(

)

Computes the frequency with which alternatives in the learning set are assigned to each category

Returns

vector of those frequency

getAlternativesPerformance()

AlternativesPerformance ProfileInitializer::getAlternativesPerformance

(

)

const

Getter of the alternativePerformance class attribute

Returns

AlternativesPerformance object

getProfilePerformanceCandidates()

std::vector<Perf> ProfileInitializer::getProfilePerformanceCandidates	(	const Criterion &	crit,
		const Category &	cat,
		const int	nbCategories
	)

Get ids of potential alternative candidates to initialize the profile performance for one specific criterion and category

Parameters

crit	Criterion object
cat	Category object
nbCategories	number of categories in our dataset

Returns

vector of alternative ids

initializeProfilePerformance()

std::vector<Perf> ProfileInitializer::initializeProfilePerformance	(	const Criterion &	crit,
		Categories &	categories,
		const std::vector< float > &	catFre
	)

Initialize all of the profile performance values for Criterion crit

Parameters

crit	Criterion object
categories	Categories object
catFrequency	category frequency of our dataset

Returns

initialized profile performances for Criterion crit

initializeProfiles()

void ProfileInitializer::initializeProfiles

(

MRSortModel &

model

)

Updates profile attribute from MRSortModel class with the a new profile given by the metaheuristic.

Parameters

MRSortModel

Mrsort model object

setAlternativesPerformance()

void ProfileInitializer::setAlternativesPerformance

(

AlternativesPerformance &

newAltPerfs

)

Setter of the alternativePerformance class attribute

Parameters

newAltPerfs

AlternativesPerformance objet that represents our dataset

weightedProbability()

float ProfileInitializer::weightedProbability	(	const Perf	perfAlt,
		const Criterion &	crit,
		const Category &	catAbove,
		const Category &	catBelow,
		const int	nbCategories,
		const std::vector< float > &	catFrequency,
		std::vector< Perf > &	candidates,
		float	delta = 0.001
	)

Compute the likelihood of choosing the performance of the alternative identified by its altId as the performance value for profile delimiting catAbove and catBelow for criterion crit.

Parameters

perfAlt	Perf object of an alternative
crit	Criterion object
catAbove	Category object category above imaginary profile performance (ie performance of altId)
catBelow	Category object category below imaginary profile performance (ie performance of altId)
nbCategories	number of categories in our dataset
catFrequency	category frequency of our dataset
candidates	potential candidates that were among our imaginary profile performance (ie performance of altId)
delta	articifial integer used to compare each candidates with the imaginary profile performanc.

Returns

probability of choosing alt_id for crit.

---

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

• /Users/ThomasVindard 1/Desktop/3A/Projet/fastPL/include/learning/ProfileInitializer.h