import classifier.*;
import classifier.gp.*;

public class BinaryClassifierFitness implements FitnessFunction{
	
	public BinaryClassifierFitness(Object kind,TrainingSetSV ts) {
		this.ts = (TrainingSetSVConcrete) ts;
		this.kind = (Long)kind;
	}
	
	public TrainingSet getTrainingSet() {
		return this.ts;
	}
	
	public void evaluate(Population pop) {
		
		LongBox input_red = new LongBox();
		double sumOfAdjustedFitnesses = 0.0;
		DataReaderTarghe drt = (DataReaderTarghe) ts.getDataReader();
		long pos = drt.getOccurencies(kind);
		long neg = ts.getNum() - pos;
		long ratio = neg/pos;

		for(int i=0; i < pop.getPopulationSize(); i++) {
			Individual ind = (Individual) pop.getIndividual(i);
			for(int j=0; j < ts.getNum(); j++) {
				long result;

				input_red = (LongBox) ts.getInput(j);
				result = ((GProgram) ind.getProgram()).eval(input_red.long_value);

				long ex_output = ((LongBox) ts.getOutput(j)).long_value;
		
				if(ex_output == kind.longValue()) {
					if (result == 0) 
						ind.setFN( ind.getFN()+1 );
				}
				else
					if (result == 1) 
						ind.setFP( ind.getFP()+1 );
			}

			ind.setStandardizedFitness(1 - Math.sqrt((Math.pow(ind.getFN()*ratio,2)+Math.pow(ind.getFP(),2))/(Math.pow(pos*ratio,2) + Math.pow(neg,2))));
			ind.setAdjustedFitness(1.0 / (ind.getStandardizedFitness() + 1.0));
			sumOfAdjustedFitnesses += ind.getAdjustedFitness();
		}
		for(int i = 0; i < pop.getPopulationSize(); i++) {
			Individual ind = (Individual) pop.getIndividual(i);
			ind.setNormalizedFitness(ind.getAdjustedFitness() / sumOfAdjustedFitnesses);
		}
	}
	
	protected Object getKind(){
		return kind;
	}
	
	private TrainingSetSVConcrete ts;
	private Long kind;
	
}