categorizer.java

决策树分类中经典算法的ID3和Ｃ４．５代码公共包！

package shared;
import java.lang.*;
import java.io.*;

/** Abstract base class for Categorizers. Number of categories must be
 * strictly positive (greater than zero). Description cannot be empty or NULL.
 * @author James Louis	2/25/2001	Ported to Java.
 * @author Chia-Hsin Li	11/29/94	Add function building probability
 * distribution array.
 * @author Ronny Kohavi	8/03/93	Initial revision
 */
abstract public class Categorizer extends Globals implements Cloneable {
    /** The Schema of data this Categorizer can categorize. **/
    private Schema schema;
    /** The total weight of instances. **/
    private double totalWeight;
    /** Number of possible categories for labelling. **/
    private int numCat;
    /** Description of this Categorizer. **/
    private String descr;
    /** The distribution of weights. **/
    private double[] distrArray;
    /** The original distribution of weights. **/
    private double[] originalDistr;
    /** Extra text for visualization purposes. **/
    private String extraVizText;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CATEGORIZER_ID_BASE = 0;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_CONST_CATEGORIZER =         CATEGORIZER_ID_BASE +  1;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_MULTITHRESH_CATEGORIZER =   CATEGORIZER_ID_BASE +  2;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_THRESHOLD_CATEGORIZER =     CATEGORIZER_ID_BASE +  3;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_ATTR_CATEGORIZER =          CATEGORIZER_ID_BASE +  4;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_RDG_CATEGORIZER =           CATEGORIZER_ID_BASE +  5;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_BAD_CATEGORIZER =           CATEGORIZER_ID_BASE +  6;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_TABLE_CATEGORIZER =         CATEGORIZER_ID_BASE +  7;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_IB_CATEGORIZER =            CATEGORIZER_ID_BASE +  8;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_LAZYDT_CATEGORIZER =        CATEGORIZER_ID_BASE +  9;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_NB_CATEGORIZER =            CATEGORIZER_ID_BASE + 10;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_PROJECT_CATEGORIZER =       CATEGORIZER_ID_BASE + 11;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_DISC_CATEGORIZER =          CATEGORIZER_ID_BASE + 12;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_ATTR_EQ_CATEGORIZER =       CATEGORIZER_ID_BASE + 13;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_DTREE_CATEGORIZER =         CATEGORIZER_ID_BASE + 14;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_BAGGING_CATEGORIZER =       CATEGORIZER_ID_BASE + 15;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_LINDISCR_CATEGORIZER =      CATEGORIZER_ID_BASE + 16;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_CASCADE_CATEGORIZER =       CATEGORIZER_ID_BASE + 17;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_STACKING_CATEGORIZER =      CATEGORIZER_ID_BASE + 18;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_ATTR_SUBSET_CATEGORIZER =   CATEGORIZER_ID_BASE + 19;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_MULTI_SPLIT_CATEGORIZER =   CATEGORIZER_ID_BASE + 20;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_ONE_R_CATEGORIZER =         CATEGORIZER_ID_BASE + 21;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_CONSTRUCT_CATEGORIZER =     CATEGORIZER_ID_BASE + 22;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_LEAF_CATEGORIZER =          CATEGORIZER_ID_BASE + 24;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_DISC_NODE_CATEGORIZER =     CATEGORIZER_ID_BASE + 25;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_MAJORITY_CATEGORIZER =      CATEGORIZER_ID_BASE + 95;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_ODT_CATEGORIZER =           CATEGORIZER_ID_BASE + 96;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_CLUSTER_CATEGORIZER =       CATEGORIZER_ID_BASE + 97;
    
    /** Class identification number.
     * @deprecated The Java instanceof operator should be used instead of
     * the numerical class identity system. **/
    public static final int CLASS_OPTION_CATEGORIZER =        CATEGORIZER_ID_BASE + 98;
    
    
    
    /** Logging options for this class. **/
    protected LogOptions logOptions = new LogOptions();
    
    /** Sets the logging level for this object.
     * @param level	The new logging level.
     */
    public void set_log_level(int level){logOptions.set_log_level(level);}
    
    /** Returns the logging level for this object.
     * @return The log level for this Object.
     */
    public int  get_log_level(){return logOptions.get_log_level();}
    
    /** Sets the stream to which logging options are displayed.
     * @param strm	The stream to which logs will be written.
     */
    public void set_log_stream(Writer strm)
    {logOptions.set_log_stream(strm);}
    
    /** Returns the stream to which logs for this object are written.
     * @return The stream to which logs for this object are written.
     */