splitscore.java

java数据挖掘算法

                splitAndLabelDist + "-->fatal_error");
            //      ASSERT(numerator != Globals.UNDEFINED_REAL);
            gain = numerator / divisor;
            cache.gainRatio = gain;
            if (gain < 0)
                Error.err("SplitScore::get_gain_ratio: negative gain: " +
                gain + "=" + numerator + '/' + divisor + "-->fatal_error");
        }
        if (gain < 0)
            Error.err("SplitScore::get_gain_ratio: negative gain: "+ gain +"-->fatal_error");
        return gain;
    }
    
    /** Returns cache.splitEntropy, first checking to see if it has yet been set.
     * This method updates the cache.
     * @return The split entropy value stored in the cache.
     */
    public double get_split_entropy() {
        valid_cache(); // Percolate validCache to the cache members.
        if (cache.splitEntropy == Globals.UNDEFINED_REAL && has_distribution(true))
            cache.splitEntropy = Entropy.entropy(get_split_dist(), total_weight());
        return cache.splitEntropy;
    }
    
    /** Sets the split score criterion.
     * @param choice The chosen split score criterion.
     * @see #mutualInfo
     * @see #normalizedMutualInfo
     * @see #rainRatio
     * @see #mutualInfoRatio
     * @see #externalScore
     */
    public void set_split_score_criterion(byte choice)
    {splitScoreCriterion = choice; }
    
    /** Clear (delete) distribution array data.
     *
     */
    public void reset() {
        //   delete cache.splitDist;
        cache.splitDist = null;
        //   delete cache.labelDist;
        cache.labelDist = null;
        //   delete splitAndLabelDist;
        splitAndLabelDist = null;
        theExternalScore = Globals.UNDEFINED_REAL;
        
        validCache = false;
        valid_cache();
    }
    
    /** Stores the cache.splitDist array.
     *
     * @param sDist The split distribution array to be cached.
     */
    public void set_split_dist(double[] sDist) {
        valid_cache(); // Percolate validCache to the cache members.
        // Only delete one object's distribution if it doesn't share its
        //   location with the one passed in--otherwise by deleting one, we'd
        //   be deleting both.
        if (cache.splitDist == sDist)
            Error.fatalErr("SplitScore::set_split_dist: got my own pointer");
        //   delete cache.splitDist;
        cache.splitDist = sDist;
        sDist = null;
        //   DBGSLOW(OK());
    }
    
    //public void set_split_and_label_dist(double[][] sAndLDist)
    /** Stores the splitAndLabelDist array.
     * @param sAndLDist The new split and label distribution.
     * @return The old split and label distribution.
     */
    public double[][] set_split_and_label_dist(double[][] sAndLDist) {
        valid_cache(); // Percolate validCache to the cache members.
        // We can't compare for equality and use our results because
        //   the caller's reference may be a reference to an automatic variable,
        //   in which case our delete would screw it up.
        
        // Only delete one object's distribution if it doesn't share its
        //   location with the one passed in--otherwise by deleting one, we'd
        //   be deleting both.
        if (splitAndLabelDist == sAndLDist)
            Error.fatalErr("SplitScore::set_split_and_label_dist: got my own pointer");
        reset();
        splitAndLabelDist = sAndLDist;
        sAndLDist = null;
        return sAndLDist;
    }
    
    private void copy_dist(SplitScore source) {
        if (this != source) {
            reset();
            copy_split_and_label_dist(source);
            validCache = false;
        }
    }
    
    private void copy_split_and_label_dist(SplitScore source) {
        // When the new distribution differs from the old, but the array is
        //   of the same size, don't delete the old splitAndLabelDist array.
        //   Reuse it.
        valid_cache(); // Percolate validCache to the cache members.
        if (splitAndLabelDist != null && source.splitAndLabelDist != null &&
        splitAndLabelDist.length == source.splitAndLabelDist.length) {
            // Keep the splitAndLabelDist, but nothing else in the cache.
            double[][] savedDistribution = splitAndLabelDist;
            splitAndLabelDist = null;
            reset();
            splitAndLabelDist = savedDistribution;
            theExternalScore = Globals.UNDEFINED_REAL;
            splitAndLabelDist = source.get_split_and_label_dist();
        }
        else {
            reset();
            if (source.splitAndLabelDist != null)
                splitAndLabelDist = Matrix.copy(source.get_split_and_label_dist());
        }
    }
    
    /** Set the external score. The score must be non-negative, although we could
     * change it to anything but UNDEFINED_REAL.
     * @param extScore The external score value.
     */
    
    public void set_external_score(double extScore) {
        if (extScore < 0)
            Error.fatalErr("SplitScore::set_external_score: score="+extScore
            +" is negative");
        
        if (!has_distribution(false))
            Error.fatalErr("SplitScore::set_external_score: no distribution");
        MLJ.ASSERT(extScore != Globals.UNDEFINED_REAL,
        "SplitScore.set_external_score: extScore == Globals.UNDEFINED_REAL.");
        theExternalScore = extScore;
    }
    
    /** Checks if an external score has been set.
     * @return TRUE if the external score is set, FALSE otherwise.
     */
    public boolean has_external_score(){return theExternalScore != Globals.UNDEFINED_REAL;}
    
    
    /** Produces formatted display of contents of object. This method updates the
     * cache. Does not abort on unset data.
     */
    public void display() {
        display(Globals.Mcout,DisplayPref.defaultDisplayPref);
    }
    
    /** Produces formatted display of contents of object. This method updates the
     * cache. Does not abort on unset data. This method updates the cache.
     * @param stream The Writer to be displayed to.
     */
    public void display(Writer stream) {
        display(stream,DisplayPref.defaultDisplayPref);
    }
    
    
    /** Produces formatted display of contents of object. This method updates the
     * cache. Does not abort on unset data. This method updates the cache.
     * @param stream The Writer to be displayed to.
     * @param dp The display preferences.
     */
    public void display(Writer stream, DisplayPref dp) {
        String endl = "\n";
        String INDENT = "  ";
        String UNSET = "unset";
        double val;
        try{
            if (dp.preference_type() == DisplayPref.ASCIIDisplay) {
                valid_cache();
                stream.write("SplitScore:" + endl);
                stream.write(INDENT + "split score criterion = "
                + splitScoreCriterionEnum[get_split_score_criterion()] + endl);
                if (!has_distribution(false))
                    stream.write(INDENT + "has no cache data" + endl);
                else {
                    stream.write(INDENT + "total weight = ");
                    if (Globals.UNDEFINED_REAL == (val = total_weight()))
                        stream.write(UNSET + endl);
                    else
                        stream.write(val + endl);
                    if (num_splits() > 0)
                        stream.write(INDENT + "number of splitting values = "
                        + num_splits() + endl);
                    stream.write(INDENT + "entropy = ");
                    if (Globals.UNDEFINED_REAL == (val = get_entropy()))
                        stream.write(UNSET + endl);
                    else
                        stream.write(val + endl);
                    stream.write(INDENT + "split entropy = ");
                    if (Globals.UNDEFINED_REAL == (val = get_split_entropy()))
                        stream.write(UNSET + endl);
                    else
                        stream.write(val + endl);
                    stream.write(INDENT + "conditional entropy = ");
                    if (Globals.UNDEFINED_REAL == (val = get_cond_entropy()))
                        stream.write(UNSET + endl);
                    else
                        stream.write(val + endl);
                    stream.write(INDENT + "mutual information = "
                    + get_mutual_info(false) + endl);
                    stream.write(INDENT + "normalized mutual information = "
                    + get_mutual_info(true) + endl);
                    stream.write(INDENT + "gain ratio = ");
                    if (Globals.UNDEFINED_REAL == (val = get_gain_ratio()))
                        stream.write(UNSET + endl);
                    else
                        stream.write(val + endl);
                    if (has_external_score())
                        stream.write(INDENT + "external score = " + get_external_score()
                        + endl);
                    // Choose not to stream the arrays at low log levels.
                    // Note that we use stream, not get_log_stream() here
                    //   because they should go to the same stream.
                    int LEVEL = 4;
                    if (has_distribution(false)) {
                        if (get_log_level() >= LEVEL) stream.write(INDENT + "splitAndLabelDist = "
                        + get_split_and_label_dist() + endl);
                        if (get_log_level() >= LEVEL) stream.write(INDENT + "labelDist = "
                        + get_label_dist() + endl);
                        if (get_log_level() >= LEVEL) stream.write(INDENT + "splitDist = "
                        + get_split_dist() + endl);
                    }
                }
            }
        }catch(IOException e){e.printStackTrace(); System.exit(1);}
    }
    
    /** Assigns the given SplitScore data to this SplitScore.
     * @param rhs The SplitScore to be copied.
     * @return This SplitScore after assignment.
     */
    public SplitScore assign(SplitScore rhs) {
        if (this != rhs) {
            copy_dist(rhs);
            splitScoreCriterion = rhs.splitScoreCriterion;
            theExternalScore = rhs.theExternalScore;
        }
        return (this);
    }
    
    /** Returns the split and label distribution array and releases ownership.
     *
     * @return The split and label distribution.
     */
    public double[][] release_split_and_label_dist() {
        double[][] distribution = splitAndLabelDist;
        if (distribution == null)
            Error.fatalErr("SplitScore::release_split_and_label_dist: there is no "
            +"distribution to release");
        splitAndLabelDist = null;
        reset();
        return distribution;
    }
    
    /** Returns the label distribution array and releases ownership.
     * @return The label distribution.
     */
    public double[] release_label_dist() {
        get_label_dist();
        double[] distribution = cache.labelDist;
        if (distribution == null)
            Error.fatalErr("SplitScore::release_label_dist: there is no "
            +"label distribution to release");
        cache.labelDist = null;
        return distribution;
    }
    
    /** Returns the split distribution array and releases ownership.
     * @return The split distribution.
     */
    public double[] release_split_dist() {
        get_split_dist();
        double[] distribution = cache.splitDist;
        if (distribution == null)
            Error.fatalErr("SplitScore::release_split_dist: there is no "
            +"split distribution to release");
        cache.splitDist = null;
        return distribution;
    }
}