de.uniba.sme.bambirds.level_selection.Decision Class Reference

Collaboration diagram for de.uniba.sme.bambirds.level_selection.Decision:

Public Member Functions
	Decision (boolean primitiveCombination)
Map< Integer, Double >	calculateProbabilityDistribution (Map< Integer, Integer > maxScores, Map< Integer, Long > costs, Map< Integer, PredictionTuple< Integer, Double > > predictions, Map< Integer, Level.State > levelStates, long timeLeft)

Private Member Functions
Map< Integer, Double >	calculateImprovementPrediction (Map< Integer, Integer > maxScores, Map< Integer, Long > costs, Map< Integer, PredictionTuple< Integer, Double > > predictions, double brierScoreLoss, double scoresError, Map< Integer, State > levelStates, long timeLeft)

Private Attributes
final boolean	primitiveCombination

Static Private Attributes
static final Logger	log = LogManager.getLogger(Decision.class)

Detailed Description

Class for evaluating probabilities of levels from predictions and level states

Constructor & Destructor Documentation

Decision()

de.uniba.sme.bambirds.level_selection.Decision.Decision

(

boolean

primitiveCombination

)

Parameters

primitiveCombination

if the combination of score prediction and win prediction should be primitive

Member Function Documentation

calculateImprovementPrediction()

Map< Integer, Double > de.uniba.sme.bambirds.level_selection.Decision.calculateImprovementPrediction ( Map< Integer, Integer >  maxScores, Map< Integer, Long >  costs, Map< Integer, PredictionTuple< Integer, Double > >  predictions, double  brierScoreLoss, double  scoresError, Map< Integer, State >  levelStates, long  timeLeft  )

private

Calculate the improvement based on the accuracy of the predictions. More info about the formula can be found in the Project Report 2019 under Level-Selection->Architecture->Decision

Parameters

maxScores	The max achieved scores of the Levels
costs	The Costs for a Levels, measured in time needed
predictions	The predictions for all Levels
brierScoreLoss	The error of the Won-Classifier
scoresError	The error of the Score-Rregressor
levelStates	The states all Levels are in. If state is Resigned, improvement will not be calculated
timeLeft	The time left in the competition

Returns

The improvement prediction, not measured in Score points, only difference important

calculateProbabilityDistribution()

Map< Integer, Double > de.uniba.sme.bambirds.level_selection.Decision.calculateProbabilityDistribution	(	Map< Integer, Integer >	maxScores,
		Map< Integer, Long >	costs,
		Map< Integer, PredictionTuple< Integer, Double > >	predictions,
		Map< Integer, Level.State >	levelStates,
		long	timeLeft
	)

Calculates the probability of selection for each level

Parameters

maxScores	The until now achieved scores. Structure: {levelid: score}
costs	The costs of all levels measured by the time needed to complete
predictions	The predictions for all levels in the next round. Structure: {levelid: PredictionTuple(scorePrediction, winProbability)}
levelStates	The states all levels are currently in. Structure: {levelid: State}
timeLeft	The remaining time for the competition

Returns

The probability distribution for all levels. Structure: {levelid: probability}

---

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

• Decision.java