/**
 * Copyright (c) 2011, The University of Southampton and the individual contributors.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modification,
 * are permitted provided that the following conditions are met:
 *
 *   *  Redistributions of source code must retain the above copyright notice,
 *      this list of conditions and the following disclaimer.
 *
 *   *  Redistributions in binary form must reproduce the above copyright notice,
 *      this list of conditions and the following disclaimer in the documentation
 *      and/or other materials provided with the distribution.
 *
 *   *  Neither the name of the University of Southampton nor the names of its
 *      contributors may be used to endorse or promote products derived from this
 *      software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
/**
 *
 */
package org.openimaj.experiment.evaluation.agreement;

import gnu.trove.map.hash.TObjectIntHashMap;

import java.util.Map;

/**
 * Calculates the interrater agreement for a given dataset between two (and only
 * two) raters.
 * <p>
 * Cohen's Kappa is defined as: (PrA - PrE) / (1 - PrE) where PrA is the
 * percentage agreement, and PrE is the probability of random agreement. PrA =
 * agreement / total and PrE = PrX + PrY where PrX and PrY are the probability
 * of both agreeing on X or both agreeing on Y randomly (that is,
 * Pr(r1,x)*Pr(r2,x) ... )
 * 
 * @see "http://en.wikipedia.org/wiki/Cohen's_kappa"
 * @author David Dupplaw (dpd@ecs.soton.ac.uk)
 * @created 12 Aug 2013
 */
public class CohensKappaInterraterAgreement
{
        /**
         * The input should be a {@link Map} for each rater where the keys represent
         * all the subjects that were rated by the raters and the values represent
         * the annotations given by the raters. Agreement between the raters is
         * determined by {@link #equals(Object)} for the INSTANCE type. Annotations
         * for subjects which are not in both sets are ignored.
         * 
         * @see "http://en.wikipedia.org/wiki/Cohen's_kappa"
         * 
         * @param rater1
         *            The annotations from rater 1
         * @param rater2
         *            The annotations from rater 2
         * @return Cohen's Kappa [0,1]
         */
        public static <K, A> double calculate(
                        final Map<K, A> rater1,
                        final Map<K, A> rater2)
        {
                int totalCount = 0;
                int agreementCount = 0;
                final TObjectIntHashMap<A> answerCountsR1 = new TObjectIntHashMap<A>();
                final TObjectIntHashMap<A> answerCountsR2 = new TObjectIntHashMap<A>();

                for (final K subjectKey : rater1.keySet())
                {
                        // We can only form an agreement if both raters rated this
                        // specific subject, so let's check
                        if (rater2.keySet().contains(subjectKey))
                        {
                                final A r1a = rater1.get(subjectKey);
                                final A r2a = rater2.get(subjectKey);

                                // It's possible that the key exists but is mapped to
                                // a null value (for example, if majority voting was used
                                // to generate the set and there was no majority).
                                if (r1a == null || r2a == null)
                                        continue;

                                // Get the answers from the raters
                                final A annotation1 = r1a;
                                final A annotation2 = r2a;

                                // Count the agreements
                                if (annotation1.equals(annotation2))
                                        agreementCount++;

                                // Count each of the answers for each of the raters
                                answerCountsR1.putIfAbsent(annotation1, 0);
                                answerCountsR2.putIfAbsent(annotation2, 0);
                                answerCountsR1.increment(annotation1);
                                answerCountsR2.increment(annotation2);

                                // Keep a running total
                                totalCount++;
                        }
                }

                System.out.println(answerCountsR1);

                final double PrA = agreementCount / (double) totalCount;
                System.out.println(PrA);

                double PrE = 0;
                for (final A ann : answerCountsR1.keySet())
                {
                        final Integer i = answerCountsR2.get(ann);
                        final double PrAnnR1 = answerCountsR1.get(ann) / (double) totalCount;
                        final double PrAnnR2 = (i == null ? 0 : i) / (double) totalCount;
                        PrE += PrAnnR1 * PrAnnR2;
                }
                System.out.println(PrE);

                final double kappa = (PrA - PrE) / (1d - PrE);

                return kappa;
        }
}