weka.knowledgeflow.steps.TimeSeriesForecasting Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of timeseriesForecasting Show documentation

Provides a time series forecasting environment for Weka. Includes a wrapper for Weka regression schemes that automates the process of creating lagged variables and date-derived periodic variables and provides the ability to do closed-loop forecasting. New evaluation routines are provided by a special evaluation module and graphing of predictions/forecasts are provided via the JFreeChart library. Includes both command-line and GUI user interfaces. Sample time series data can be found in ${WEKA_HOME}/packages/timeseriesForecasting/sample-data.

There is a newer version: 1.1.27

Show newest version

/*
 *   This program is free software: you can redistribute it and/or modify
 *   it under the terms of the GNU General Public License as published by
 *   the Free Software Foundation, either version 3 of the License, or
 *   (at your option) any later version.
 *
 *   This program is distributed in the hope that it will be useful,
 *   but WITHOUT ANY WARRANTY; without even the implied warranty of
 *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *   GNU General Public License for more details.
 *
 *   You should have received a copy of the GNU General Public License
 *   along with this program.  If not, see .
 */


/*
 *    TimeSeriesForecasting.java
 *    Copyright (C) 2010-2016 University of Waikato, Hamilton, New Zealand
 */

package weka.knowledgeflow.steps;

import org.apache.commons.codec.binary.Base64;
import weka.classifiers.evaluation.NumericPrediction;
import weka.classifiers.timeseries.AbstractForecaster;
import weka.classifiers.timeseries.WekaForecaster;
import weka.core.SerializationHelper;
import weka.filters.supervised.attribute.TSLagMaker;
import weka.core.Attribute;
import weka.core.DenseInstance;
import weka.core.Instance;
import weka.core.Instances;
import weka.core.OptionMetadata;
import weka.core.Utils;
import weka.core.WekaException;
import weka.gui.FilePropertyMetadata;
import weka.gui.ProgrammaticProperty;
import weka.gui.knowledgeflow.KFGUIConsts;
import weka.knowledgeflow.Data;
import weka.knowledgeflow.StepManager;

import javax.swing.JFileChooser;
import java.io.BufferedInputStream;
import java.io.BufferedOutputStream;
import java.io.ByteArrayInputStream;
import java.io.ByteArrayOutputStream;
import java.io.File;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.InputStream;
import java.io.ObjectInputStream;
import java.io.ObjectOutputStream;
import java.io.OutputStream;
import java.util.ArrayList;
import java.util.List;
import java.util.zip.GZIPInputStream;
import java.util.zip.GZIPOutputStream;

/**
 * Knowledge Flow step that encapsulates a time series forecasting model and
 * uses it to produce forecasts given incoming historical data.
 *
 * @author Mark Hall
 * @version $Revision: $
 */
@KFStep(name = "TimeSeriesForecasting", category = "TimeSeries",
  toolTipText = "Encapsulates a time series forecasting model and uses it to"
    + " produce forecasts given incoming historical data. Forecaster "
    + "can optionally be rebuilt using the incoming data before a "
    + "forecast is generated.",
  iconPath = KFGUIConsts.BASE_ICON_PATH + "DefaultClassifier.gif")
public class TimeSeriesForecasting extends BaseStep {

  /**
   * For serialization
   */
  private static final long serialVersionUID = -7826178727365267059L;

  /** The structure of the data used to train the forecaster */
  protected transient Instances m_header;

  /** The forecaster to use for forecasting */
  protected transient WekaForecaster m_forecaster;

  /**
   * The output instances structure - typically the same as the input structure.
   * Will have additional attributes for upper and lower confidence intervals if
   * the forecaster produces them
   */
  protected transient Instances m_outgoingStructure;

  /**
   * The filename to load from - takes precendence over an encoded forecaster if
   * not null and not equal to "-NONE-"
   */
  protected File m_fileName = new File("");

  /**
   * The file name to save the updated forecaster to if the user has opted to
   * rebuild the forecasting model on the incoming data
   */
  protected File m_saveFileName = new File("");

  /**
   * Base 64 encoded forecasting model - this allows the model to be embeded in
   * the XML knowledge flow file format rather than loaded from a file at
   * execution time.
   */
  protected String m_encodedForecaster = "-NONE-";

  /**
   * Number of future time steps to forecast - will be ignored if overlay data
   * is being used since the number of instances containing overlay values will
   * dictate the number of forecasted values that can be produced
   */
  protected String m_numberOfStepsToForecast = "1";

  /** True if the forecaster should be rebuilt on incoming data */
  protected boolean m_rebuildForecaster;

  /**
   * The number of time units beyond the end of the training data used to train
   * the forecaster that the most recent incoming priming instance is. This is
   * used to adjust the artificial time stamp (if one is being used) to the
   * right value before a forecast is produced
   */
  protected String m_artificialTimeStartOffset = "0";

  /** holds overlay data (if present in the incoming data) */
  protected transient Instances m_overlayData;

  /**
   * holds the incoming data for either priming the forecaster or for rebuilding
   * the forecaster
   */
  protected transient Instances m_bufferedPrimeData;

  /** true if the forecaster is using overlay data */
  protected transient boolean m_isUsingOverlayData;

  /** the lag maker in use by the forecaster */
  protected transient TSLagMaker m_modelLagMaker;

  /** name of the time stamp attribute */
  protected transient String m_timeStampName = "";

  /** the fields that the forecaster is predicting */
  protected transient List m_fieldsToForecast;

  /** True if the step has been reset */
  protected boolean m_isReset;

  /** True if the step has incoming streaming data */
  protected boolean m_isStreaming;

  /** Reusable data object for outputting forecast data */
  protected Data m_streamingData;

  /**
   * Set the base 64 encoded forecaster.
   *
   * @param encodedForecaster a base 64 encoded List