/*
 * Created on Jul 25, 2005
 *
 */
package aima.test.learningtest;

import java.util.Hashtable;

import junit.framework.TestCase;
import aima.learning.framework.DataSet;
import aima.learning.framework.DataSetFactory;
import aima.util.Util;

public class InformationAndGainTest extends TestCase {
	public void testInformationCalculation(){
		double[] fairCoinProbabilities = new double[]{0.5,0.5};
		double[] loadedCoinProbabilities = new double[]{0.01,0.99};
		
		assertEquals(1.0,Util.information(fairCoinProbabilities));
		assertEquals(0.08079313589591118,Util.information(loadedCoinProbabilities));
	}
	
	public void testBasicDataSetInformationCalculation() throws Exception{
		DataSet ds = DataSetFactory.getRestaurantDataSet();
		double infoForTargetAttribute = ds.getInformationFor();//this should be the generic distribution
		assertEquals(1.0,infoForTargetAttribute);
	}
	
	public void testDataSetSplit() throws Exception{
		DataSet ds = DataSetFactory.getRestaurantDataSet();
		Hashtable<String,DataSet> hash = ds.splitByAttribute("patrons");//this should be the generic distribution
		assertEquals(3,hash.keySet().size());
		assertEquals(6,hash.get("Full").size());
		assertEquals(2,hash.get("None").size());
		assertEquals(4,hash.get("Some").size());
		
	}
	
	public void testGainCalculation() throws Exception{
		DataSet ds = DataSetFactory.getRestaurantDataSet();
		Hashtable<String,DataSet> hash = ds.splitByAttribute("patrons");
		double gain = ds.calculateGainFor("patrons");
		assertEquals(0.541,gain,0.001);
		gain = ds.calculateGainFor("type");
		assertEquals(0.0,gain,0.001);
	}

}