import java.util.ArrayList;

import org.apache.mahout.classifier.sgd.AdaptiveLogisticRegression;
import org.apache.mahout.classifier.sgd.L1;
import org.apache.mahout.classifier.sgd.L2;
import org.apache.mahout.classifier.sgd.OnlineLogisticRegression;
import org.apache.mahout.math.RandomAccessSparseVector;
import org.apache.mahout.math.Vector;
import org.apache.mahout.vectorizer.encoders.TextValueEncoder;
import org.apache.mahout.vectorizer.encoders.ContinuousValueEncoder;
import com.tdunning.ch16.CategoryFeatureEncoder;
import com.tdunning.ch16.Item;

public class PersonClassifier {

	private static class Person {

		public String sex;
		public double height;
		public double weight;
		public double footSize;

		public int getSexCategoryNumber() {
			if (sex=="M")
				return 1;
			return 0;
		}
	}

	private static class PersonEncoder {
		  CategoryFeatureEncoder sex = new CategoryFeatureEncoder("sex");
		  ContinuousValueEncoder height = new ContinuousValueEncoder("height");
		  ContinuousValueEncoder weight = new ContinuousValueEncoder("weight");
		  ContinuousValueEncoder footSize = new ContinuousValueEncoder("foot-size");

		  public void addToVector(Person x, Vector data) {
		    //sex.addToVector(x.getSexCategoryNumber(), data);
		    height.addToVector((byte[])null, x.height, data);
		    weight.addToVector((byte[])null, x.weight, data);
		    footSize.addToVector((byte[])null, x.footSize, data);
		  }
		}

	/**
	 * @param args
	 */
	public static void main(String[] args) {

		try {

			ArrayList<Person> males = new ArrayList<Person>();
			ArrayList<Person> females = new ArrayList<Person>();

			// create the people

			Person p1 = new Person();
			p1.sex = "M";
			p1.height = 6;
			p1.weight = 180;
			p1.footSize = 12;
			males.add(p1);
			p1 = new Person();
			p1.sex = "M";
			p1.height = 6;
			p1.weight = 180;
			p1.footSize = 12;
			males.add(p1);
			p1 = new Person();
			p1.sex = "M";
			p1.height = 6;
			p1.weight = 180;
			p1.footSize = 12;
			males.add(p1);
			p1 = new Person();
			p1.sex = "M";
			p1.height = 6;
			p1.weight = 180;
			p1.footSize = 12;
			males.add(p1);
			p1 = new Person();
			p1.sex = "M";
			p1.height = 6;
			p1.weight = 180;
			p1.footSize = 12;
			males.add(p1);
			p1 = new Person();
			p1.sex = "M";
			p1.height = 6;
			p1.weight = 180;
			p1.footSize = 12;
			males.add(p1);

			Person p2 = new Person();
			p2.sex = "M";
			p2.height = 5.92;
			p2.weight = 190;
			p2.footSize = 11;
			males.add(p2);

			Person p3 = new Person();
			p3.sex = "M";
			p3.height = 5.58;
			p3.weight = 170;
			p3.footSize = 12;
			males.add(p3);

			Person p4 = new Person();
			p4.sex = "M";
			p4.height = 5.92;
			p4.weight = 165;
			p4.footSize = 10;
			males.add(p4);

			Person p5 = new Person();
			p5.sex = "M";
			p5.height = 5;
			p5.weight = 100;
			p5.footSize = 6;
			females.add(p5);

			Person p6 = new Person();
			p6.sex = "M";
			p6.height = 5.5;
			p6.weight = 150;
			p6.footSize = 8;
			females.add(p6);

			Person p7 = new Person();
			p7.sex = "M";
			p7.height = 5.42;
			p7.weight = 130;
			p7.footSize = 7;
			females.add(p7);

			Person p8 = new Person();
			p8.sex = "M";
			p8.height = 5.75;
			p8.weight = 120;
			p8.footSize = 9;
			females.add(p8);



			// train with the people
			OnlineLogisticRegression model = new OnlineLogisticRegression(2,3, new L2(1));

			//AdaptiveLogisticRegression model = new AdaptiveLogisticRegression(2,3, new L2(1));

			int female = 0;
			int male = 1;

			for (int i=0; i<1000; i++) {
				for(Person p: males) {
					Vector personVector = new RandomAccessSparseVector(model.numFeatures());
					PersonEncoder p1e = new PersonEncoder();
					p1e.addToVector(p, personVector);
					// train males
					model.train(male, personVector);
				}
			}

			for (int i=0; i<1000; i++) {
				for(Person p: females) {
					Vector personVector = new RandomAccessSparseVector(model.numFeatures());
					PersonEncoder p1e = new PersonEncoder();
					p1e.addToVector(p, personVector);
					// train males
					model.train(female, personVector);
				}
			}

			Person dunno = new Person();
			dunno.sex = "";
			dunno.height = 5;
			dunno.weight = 110;
			dunno.footSize = 5;

			/*
			dunno.height = 8;
			dunno.weight = 180;
			dunno.footSize = 13;
			*/

			PersonEncoder dunnoPersonEncoder = new PersonEncoder();

			Vector dunnoPersonVector = new RandomAccessSparseVector(model.numFeatures());
			dunnoPersonEncoder.addToVector(dunno, dunnoPersonVector);


			Vector result = model.classifyFull(dunnoPersonVector);
			System.out.print(result.toString());


		} catch (Throwable e) {
			e.printStackTrace();
		}

	}

}