001/**
002 * Copyright (c) 2011, The University of Southampton and the individual contributors.
003 * All rights reserved.
004 *
005 * Redistribution and use in source and binary forms, with or without modification,
006 * are permitted provided that the following conditions are met:
007 *
008 *   *  Redistributions of source code must retain the above copyright notice,
009 *      this list of conditions and the following disclaimer.
010 *
011 *   *  Redistributions in binary form must reproduce the above copyright notice,
012 *      this list of conditions and the following disclaimer in the documentation
013 *      and/or other materials provided with the distribution.
014 *
015 *   *  Neither the name of the University of Southampton nor the names of its
016 *      contributors may be used to endorse or promote products derived from this
017 *      software without specific prior written permission.
018 *
019 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
020 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
021 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
022 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
023 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
024 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
025 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
026 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
027 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
028 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
029 */
030package org.openimaj.demos.sandbox.ml.linear.learner.stream.experiments;
031
032import java.io.IOException;
033import java.net.MalformedURLException;
034import java.util.Arrays;
035import java.util.List;
036import java.util.Map;
037
038import org.openimaj.demos.sandbox.ml.linear.learner.stream.IncrementalLearnerFunction;
039import org.openimaj.demos.sandbox.ml.linear.learner.stream.IncrementalLearnerWorldSelectingEvaluator;
040import org.openimaj.demos.sandbox.ml.linear.learner.stream.ModelStats;
041import org.openimaj.demos.sandbox.ml.linear.learner.stream.StockPriceAggregator;
042import org.openimaj.demos.sandbox.ml.linear.learner.stream.twitter.USMFStatusBagOfWords;
043import org.openimaj.demos.sandbox.ml.linear.learner.stream.twitter.USMFTickMongoDBQueryStream;
044import org.openimaj.ml.linear.evaluation.SumLossEvaluator;
045import org.openimaj.ml.linear.learner.BilinearLearnerParameters;
046import org.openimaj.ml.linear.learner.init.FirstValueInitStrat;
047import org.openimaj.ml.linear.learner.init.SingleValueInitStrat;
048import org.openimaj.ml.linear.learner.init.SparseZerosInitStrategy;
049import org.openimaj.tools.twitter.modes.preprocessing.StopwordMode;
050import org.openimaj.twitter.USMFStatus;
051import org.openimaj.util.data.Context;
052import org.openimaj.util.function.Operation;
053import org.openimaj.util.function.context.ContextFunctionAdaptor;
054import org.openimaj.util.stream.window.WindowAverage;
055
056import com.mongodb.ServerAddress;
057
058/**
059 * @author Sina Samangooei (ss@ecs.soton.ac.uk)
060 *
061 */
062public class FinancialMongoStreamLearningExperiment {
063        /**
064         * @param args
065         * @throws MalformedURLException
066         * @throws IOException
067         */
068        public static void main(String[] args) throws MalformedURLException, IOException {
069                BilinearLearnerParameters params = new BilinearLearnerParameters();
070                params.put(BilinearLearnerParameters.ETA0_U, 0.02);
071                params.put(BilinearLearnerParameters.ETA0_W, 0.02);
072                params.put(BilinearLearnerParameters.LAMBDA, 0.001);
073                params.put(BilinearLearnerParameters.BICONVEX_TOL, 0.01);
074                params.put(BilinearLearnerParameters.BICONVEX_MAXITER, 10);
075                params.put(BilinearLearnerParameters.BIAS, true);
076                params.put(BilinearLearnerParameters.ETA0_BIAS, 0.5);
077                params.put(BilinearLearnerParameters.WINITSTRAT, new SingleValueInitStrat(0.1));
078                params.put(BilinearLearnerParameters.UINITSTRAT, new SparseZerosInitStrategy());
079                params.put(BilinearLearnerParameters.EXPANDEDUINITSTRAT, new SparseZerosInitStrategy());
080                params.put(BilinearLearnerParameters.EXPANDEDWINITSTRAT, new SingleValueInitStrat(0.05));
081                FirstValueInitStrat biasInitStrat = new FirstValueInitStrat();
082                params.put(BilinearLearnerParameters.BIASINITSTRAT, biasInitStrat);
083
084                List<ServerAddress> serverList = Arrays.asList(
085                                new ServerAddress("rumi", 27017),
086                                new ServerAddress("hafez", 27017)
087                                );
088                // Get the USMF status objects and financial ticks from the mongodb
089                new USMFTickMongoDBQueryStream(serverList)
090                // Transform the usmf status instances to bags of words
091                .map(
092                        new ContextFunctionAdaptor<List<USMFStatus>, Map<String,Map<String,Double>>>(
093                                new USMFStatusBagOfWords(new StopwordMode()),"usmfstatuses",
094                                "bagofwords"
095                        )
096                )
097                // transform the financial ticks to the average tick
098                .map(
099                        new ContextFunctionAdaptor<List<Map<String,Double>>,Map<String,Double>>(new WindowAverage(),"ticks","averageticks")
100                )
101                // Group together identical stock ticks
102                .transform(new StockPriceAggregator(0.0001))
103                // Train the model
104                .map(
105                        new IncrementalLearnerWorldSelectingEvaluator(
106                                new SumLossEvaluator(),
107                                new IncrementalLearnerFunction(params)
108                        )
109                )
110                // Consume the model statistics
111                .forEach(new Operation<Context>() {
112
113                        @Override
114                        public void perform(Context context) {
115                                ModelStats object = context.getTyped("modelstats");
116                                object.printSummary();
117                        }
118                });
119
120        }
121}