/*
* 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 2 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, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
/*
* ResultMatrix.java
* Copyright (C) 2005 University of Waikato, Hamilton, New Zealand
*
*/
package weka.experiment;
import weka.core.RevisionHandler;
import weka.core.Utils;
import java.io.Serializable;
import java.util.Enumeration;
import java.util.Vector;
/**
* This matrix is a container for the datasets and classifier setups and
* their statistics. Derived classes output the data in different formats.
* Derived classes need to implement the following methods:
* <ul>
* <li><code>toStringMatrix()</code></li>
* <li><code>toStringKey()</code></li>
* <li><code>toStringHeader()</code></li>
* <li><code>toStringSummary()</code></li>
* <li><code>toStringRanking()</code></li>
* </ul>
*
*
* @author FracPete (fracpete at waikato dot ac dot nz)
* @version $Revision: 1.9 $
* @see #toStringMatrix()
* @see #toStringKey()
* @see #toStringHeader()
* @see #toStringSummary()
* @see #toStringRanking()
*/
public abstract class ResultMatrix
implements Serializable, RevisionHandler {
/** for serialization */
private static final long serialVersionUID = 4487179306428209739L;
/** tie */
public final static int SIGNIFICANCE_TIE = 0;
/** win */
public final static int SIGNIFICANCE_WIN = 1;
/** loss */
public final static int SIGNIFICANCE_LOSS = 2;
/** tie string */
public String TIE_STRING = " ";
/** win string */
public String WIN_STRING = "v";
/** loss string */
public String LOSS_STRING = "*";
/** the left parentheses for enumerating cols/rows */
public String LEFT_PARENTHESES = "(";
/** the right parentheses for enumerating cols/rows */
public String RIGHT_PARENTHESES = ")";
/** the column names */
protected String[] m_ColNames = null;
/** the row names */
protected String[] m_RowNames = null;
/** whether a column is hidden */
protected boolean[] m_ColHidden = null;
/** whether a row is hidden */
protected boolean[] m_RowHidden = null;
/** the significance */
protected int[][] m_Significance = null;
/** the values */
protected double[][] m_Mean = null;
/** the standard deviation */
protected double[][] m_StdDev = null;
/** the counts for the different datasets */
protected double[] m_Counts = null;
/** the standard mean precision */
protected int m_MeanPrec;
/** the standard std. deviation preicision */
protected int m_StdDevPrec;
/** whether std. deviations are printed as well */
protected boolean m_ShowStdDev;
/** whether the average for each column should be printed */
protected boolean m_ShowAverage;
/** whether the names or numbers are output as column declarations */
protected boolean m_PrintColNames;
/** whether the names or numbers are output as row declarations */
protected boolean m_PrintRowNames;
/** whether a "(x)" is printed before each column name with "x" as the
* index */
protected boolean m_EnumerateColNames;
/** whether a "(x)" is printed before each row name with "x" as the index */
protected boolean m_EnumerateRowNames;
/** the size of the names of the columns */
protected int m_ColNameWidth;
/** the size of the names of the rows */
protected int m_RowNameWidth;
/** the size of the mean columns */
protected int m_MeanWidth;
/** the size of the std dev columns */
protected int m_StdDevWidth;
/** the size of the significance columns */
protected int m_SignificanceWidth;
/** the size of the counts */
protected int m_CountWidth;
/** contains the keys for the header */
protected Vector m_HeaderKeys = null;
/** contains the values for the header */
protected Vector m_HeaderValues = null;
/** the non-significant wins */
protected int[][] m_NonSigWins = null;
/** the significant wins */
protected int[][] m_Wins = null;
/** the wins in ranking */
protected int[] m_RankingWins = null;
/** the losses in ranking */
protected int[] m_RankingLosses = null;
/** the difference between wins and losses */
protected int[] m_RankingDiff = null;
/** the ordering of the rows */
protected int[] m_RowOrder = null;
/** the ordering of the columns */
protected int[] m_ColOrder = null;
/** whether to remove the filter name from the dataaset name */
protected boolean m_RemoveFilterName = false;
/**
* initializes the matrix as 1x1 matrix
*/
public ResultMatrix() {
this(1, 1);
}
/**
* initializes the matrix with the given dimensions
*/
public ResultMatrix(int cols, int rows) {
setSize(cols, rows);
clear();
}
/**
* initializes the matrix with the values from the given matrix
* @param matrix the matrix to get the values from
*/
public ResultMatrix(ResultMatrix matrix) {
assign(matrix);
}
/**
* returns the name of the output format
*/
public abstract String getDisplayName();
/**
* acquires the data from the given matrix
*/
public void assign(ResultMatrix matrix) {
int i;
int n;
setSize(matrix.getColCount(), matrix.getRowCount());
// output parameters
TIE_STRING = matrix.TIE_STRING;
WIN_STRING = matrix.WIN_STRING;
LOSS_STRING = matrix.LOSS_STRING;
LEFT_PARENTHESES = matrix.LEFT_PARENTHESES;
RIGHT_PARENTHESES = matrix.RIGHT_PARENTHESES;
m_MeanPrec = matrix.m_MeanPrec;
m_StdDevPrec = matrix.m_StdDevPrec;
m_ShowStdDev = matrix.m_ShowStdDev;
m_ShowAverage = matrix.m_ShowAverage;
m_PrintColNames = matrix.m_PrintColNames;
m_PrintRowNames = matrix.m_PrintRowNames;
m_EnumerateColNames = matrix.m_EnumerateColNames;
m_EnumerateRowNames = matrix.m_EnumerateRowNames;
m_RowNameWidth = matrix.m_RowNameWidth;
m_MeanWidth = matrix.m_MeanWidth;
m_StdDevWidth = matrix.m_StdDevWidth;
m_SignificanceWidth = matrix.m_SignificanceWidth;
m_CountWidth = matrix.m_CountWidth;
m_RemoveFilterName = matrix.m_RemoveFilterName;
// header
m_HeaderKeys = (Vector) matrix.m_HeaderKeys.clone();
m_HeaderValues = (Vector) matrix.m_HeaderValues.clone();
// matrix
for (i = 0; i < matrix.m_Mean.length; i++) {
for (n = 0; n < matrix.m_Mean[i].length; n++) {
m_Mean[i][n] = matrix.m_Mean[i][n];
m_StdDev[i][n] = matrix.m_StdDev[i][n];
m_Significance[i][n] = matrix.m_Significance[i][n];
}
}
for (i = 0; i < matrix.m_ColNames.length; i++) {
m_ColNames[i] = matrix.m_ColNames[i];
m_ColHidden[i] = matrix.m_ColHidden[i];
}
for (i = 0; i < matrix.m_RowNames.length; i++) {
m_RowNames[i] = matrix.m_RowNames[i];
m_RowHidden[i] = matrix.m_RowHidden[i];
}
for (i = 0; i < matrix.m_Counts.length; i++)
m_Counts[i] = matrix.m_Counts[i];
// summary
if (matrix.m_NonSigWins != null) {
m_NonSigWins = new int[matrix.m_NonSigWins.length][];
m_Wins = new int[matrix.m_NonSigWins.length][];
for (i = 0; i < matrix.m_NonSigWins.length; i++) {
m_NonSigWins[i] = new int[matrix.m_NonSigWins[i].length];
m_Wins[i] = new int[matrix.m_NonSigWins[i].length];
for (n = 0; n < matrix.m_NonSigWins[i].length; n++) {
m_NonSigWins[i][n] = matrix.m_NonSigWins[i][n];
m_Wins[i][n] = matrix.m_Wins[i][n];
}
}
}
// ranking
if (matrix.m_RankingWins != null) {
m_RankingWins = new int[matrix.m_RankingWins.length];
m_RankingLosses = new int[matrix.m_RankingWins.length];
m_RankingDiff = new int[matrix.m_RankingWins.length];
for (i = 0; i < matrix.m_RankingWins.length; i++) {
m_RankingWins[i] = matrix.m_RankingWins[i];
m_RankingLosses[i] = matrix.m_RankingLosses[i];
m_RankingDiff[i] = matrix.m_RankingDiff[i];
}
}
}
/**
* removes the stored data and the ordering, but retains the dimensions of
* the matrix
*/
public void clear() {
m_MeanPrec = 2;
m_StdDevPrec = 2;
m_ShowStdDev = false;
m_ShowAverage = false;
m_PrintColNames = true;
m_PrintRowNames = true;
m_EnumerateColNames = true;
m_EnumerateRowNames = false;
m_RowNameWidth = 0;
m_ColNameWidth = 0;
m_MeanWidth = 0;
m_StdDevWidth = 0;
m_SignificanceWidth = 0;
m_CountWidth = 0;
setSize(getColCount(), getRowCount());
}
/**
* clears the content of the matrix and sets the new size
* @param cols the number of mean columns
* @param rows the number of mean rows
*/
public void setSize(int cols, int rows) {
int i;
int n;
m_ColNames = new String[cols];
m_RowNames = new String[rows];
m_Counts = new double[rows];
m_ColHidden = new boolean[cols];
m_RowHidden = new boolean[rows];
m_Mean = new double[rows][cols];
m_Significance = new int[rows][cols];
m_StdDev = new double[rows][cols];
m_ColOrder = null;
m_RowOrder = null;
// NaN means that there exists no value! -> toArray()
for (i = 0; i < m_Mean.length; i++) {
for (n = 0; n < m_Mean[i].length; n++)
m_Mean[i][n] = Double.NaN;
}
for (i = 0; i < m_ColNames.length; i++)
m_ColNames[i] = "col" + i;
for (i = 0; i < m_RowNames.length; i++)
m_RowNames[i] = "row" + i;
clearHeader();
clearSummary();
clearRanking();
}
/**
* sets the precision for the means
*/
public void setMeanPrec(int prec) {
if (prec >= 0)
m_MeanPrec = prec;
}
/**
* returns the current precision for the means
*/
public int getMeanPrec() {
return m_MeanPrec;
}
/**
* sets the precision for the standard deviation
*/
public void setStdDevPrec(int prec) {
if (prec >= 0)
m_StdDevPrec = prec;
}
/**
* returns the current standard deviation precision
*/
public int getStdDevPrec() {
return m_StdDevPrec;
}
/**
* sets the width for the column names (0 = optimal)
*/
public void setColNameWidth(int width) {
if (width >= 0)
m_ColNameWidth = width;
}
/**
* returns the current width for the column names
*/
public int getColNameWidth() {
return m_ColNameWidth;
}
/**
* sets the width for the row names (0 = optimal)
*/
public void setRowNameWidth(int width) {
if (width >= 0)
m_RowNameWidth = width;
}
/**
* returns the current width for the row names
*/
public int getRowNameWidth() {
return m_RowNameWidth;
}
/**
* sets the width for the mean (0 = optimal)
*/
public void setMeanWidth(int width) {
if (width >= 0)
m_MeanWidth = width;
}
/**
* returns the current width for the mean
*/
public int getMeanWidth() {
return m_MeanWidth;
}
/**
* sets the width for the std dev (0 = optimal)
*/
public void setStdDevWidth(int width) {
if (width >= 0)
m_StdDevWidth = width;
}
/**
* returns the current width for the std dev
*/
public int getStdDevWidth() {
return m_StdDevWidth;
}
/**
* sets the width for the significance (0 = optimal)
*/
public void setSignificanceWidth(int width) {
if (width >= 0)
m_SignificanceWidth = width;
}
/**
* returns the current width for the significance
*/
public int getSignificanceWidth() {
return m_SignificanceWidth;
}
/**
* sets the width for the counts (0 = optimal)
*/
public void setCountWidth(int width) {
if (width >= 0)
m_CountWidth = width;
}
/**
* returns the current width for the counts
*/
public int getCountWidth() {
return m_CountWidth;
}
/**
* sets whether to display the std deviations or not
*/
public void setShowStdDev(boolean show) {
m_ShowStdDev = show;
}
/**
* returns whether std deviations are displayed or not
*/
public boolean getShowStdDev() {
return m_ShowStdDev;
}
/**
* sets whether to display the average per column or not
*/
public void setShowAverage(boolean show) {
m_ShowAverage = show;
}
/**
* returns whether average per column is displayed or not
*/
public boolean getShowAverage() {
return m_ShowAverage;
}
/**
* sets whether to remove the filter classname from the dataset name
*/
public void setRemoveFilterName(boolean remove) {
m_RemoveFilterName = remove;
}
/**
* returns whether the filter classname is removed from the dataset name
*/
public boolean getRemoveFilterName() {
return m_RemoveFilterName;
}
/**
* sets whether the column names or numbers instead are printed.
* deactivating automatically sets m_EnumerateColNames to TRUE.
* @see #setEnumerateColNames(boolean)
*/
public void setPrintColNames(boolean print) {
m_PrintColNames = print;
if (!print)
setEnumerateColNames(true);
}
/**
* returns whether column names or numbers instead are printed
*/
public boolean getPrintColNames() {
return m_PrintColNames;
}
/**
* sets whether the row names or numbers instead are printed
* deactivating automatically sets m_EnumerateColNames to TRUE.
* @see #setEnumerateRowNames(boolean)
*/
public void setPrintRowNames(boolean print) {
m_PrintRowNames = print;
if (!print)
setEnumerateRowNames(true);
}
/**
* returns whether row names or numbers instead are printed
*/
public boolean getPrintRowNames() {
return m_PrintRowNames;
}
/**
* sets whether the column names are prefixed with "(x)" where "x" is
* the index
*/
public void setEnumerateColNames(boolean enumerate) {
m_EnumerateColNames = enumerate;
}
/**
* returns whether column names or numbers instead are enumerateed
*/
public boolean getEnumerateColNames() {
return m_EnumerateColNames;
}
/**
* sets whether to the row names or numbers instead are enumerateed
*/
public void setEnumerateRowNames(boolean enumerate) {
m_EnumerateRowNames = enumerate;
}
/**
* returns whether row names or numbers instead are enumerateed
*/
public boolean getEnumerateRowNames() {
return m_EnumerateRowNames;
}
/**
* returns the number of columns
*/
public int getColCount() {
return m_ColNames.length;
}
/**
* returns the number of visible columns
*/
public int getVisibleColCount() {
int cols;
int i;
cols = 0;
for (i = 0; i < getColCount(); i++) {
if (!getColHidden(i))
cols++;
}
return cols;
}
/**
* returns the number of rows
*/
public int getRowCount() {
return m_RowNames.length;
}
/**
* returns the number of visible rows
*/
public int getVisibleRowCount() {
int rows;
int i;
rows= 0;
for (i = 0; i < getRowCount(); i++) {
if (!getRowHidden(i))
rows++;
}
return rows;
}
/**
* sets the name of the column (if the index is valid)
* @param index the index of the column
* @param name the name of the column
*/
public void setColName(int index, String name) {
if ( (index >= 0) && (index < getColCount()) )
m_ColNames[index] = name;
}
/**
* returns the name of the row, if the index is valid, otherwise null.
* if getPrintColNames() is FALSE then an empty string is returned or if
* getEnumerateColNames() is TRUE then the 1-based index surrounded by
* parentheses.
* @see #setPrintColNames(boolean)
* @see #getPrintColNames()
* @see #setEnumerateColNames(boolean)
* @see #getEnumerateColNames()
*/
public String getColName(int index) {
String result;
result = null;
if ( (index >= 0) && (index < getColCount()) ) {
if (getPrintColNames())
result = m_ColNames[index];
else
result = "";
if (getEnumerateColNames()) {
result = LEFT_PARENTHESES
+ Integer.toString(index + 1)
+ RIGHT_PARENTHESES
+ " " + result;
result = result.trim();
}
}
return result;
}
/**
* sets the name of the row (if the index is valid)
* @param index the index of the row
* @param name the name of the row
*/
public void setRowName(int index, String name) {
if ( (index >= 0) && (index < getRowCount()) )
m_RowNames[index] = name;
}
/**
* returns the name of the row, if the index is valid, otherwise null.
* if getPrintRowNames() is FALSE then an empty string is returned or if
* getEnumerateRowNames() is TRUE then the 1-based index surrounded by
* parentheses.
* @see #setPrintRowNames(boolean)
* @see #getPrintRowNames()
* @see #setEnumerateRowNames(boolean)
* @see #getEnumerateRowNames()
*/
public String getRowName(int index) {
String result;
result = null;
if ( (index >= 0) && (index < getRowCount()) ) {
if (getPrintRowNames())
result = m_RowNames[index];
else
result = "";
if (getEnumerateRowNames()) {
result = LEFT_PARENTHESES
+ Integer.toString(index + 1)
+ RIGHT_PARENTHESES
+ " " + result;
result = result.trim();
}
}
return result;
}
/**
* sets the hidden status of the column (if the index is valid)
* @param index the index of the column
* @param hidden the hidden status of the column
*/
public void setColHidden(int index, boolean hidden) {
if ( (index >= 0) && (index < getColCount()) )
m_ColHidden[index] = hidden;
}
/**
* returns the hidden status of the column, if the index is valid, otherwise
* false
*/
public boolean getColHidden(int index) {
if ( (index >= 0) && (index < getColCount()) )
return m_ColHidden[index];
else
return false;
}
/**
* sets the hidden status of the row (if the index is valid)
* @param index the index of the row
* @param hidden the hidden status of the row
*/
public void setRowHidden(int index, boolean hidden) {
if ( (index >= 0) && (index < getRowCount()) )
m_RowHidden[index] = hidden;
}
/**
* returns the hidden status of the row, if the index is valid, otherwise
* false
*/
public boolean getRowHidden(int index) {
if ( (index >= 0) && (index < getRowCount()) )
return m_RowHidden[index];
else
return false;
}
/**
* sets the count for the row (if the index is valid)
* @param index the index of the row
* @param count the count for the row
*/
public void setCount(int index, double count) {
if ( (index >= 0) && (index < getRowCount()) )
m_Counts[index] = count;
}
/**
* returns the count for the row. if the index is invalid then 0.
* @param index the index of the row
* @return the count for the row
*/
public double getCount(int index) {
if ( (index >= 0) && (index < getRowCount()) )
return m_Counts[index];
else
return 0;
}
/**
* sets the mean at the given position (if the position is valid)
* @param col the column of the mean
* @param row the row of the mean
* @param value the value of the mean
*/
public void setMean(int col, int row, double value) {
if ( (col >= 0) && (col < getColCount())
&& (row >= 0) && (row < getRowCount()) )
m_Mean[row][col] = value;
}
/**
* returns the mean at the given position, if the position is valid,
* otherwise 0
*/
public double getMean(int col, int row) {
if ( (col >= 0) && (col < getColCount())
&& (row >= 0) && (row < getRowCount()) )
return m_Mean[row][col];
else
return 0;
}
/**
* returns the average of the mean at the given position, if the position is
* valid, otherwise 0
*/
public double getAverage(int col) {
int i;
double avg;
int count;
if ( (col >= 0) && (col < getColCount()) ) {
avg = 0;
count = 0;
for (i = 0; i < getRowCount(); i++) {
if (!Double.isNaN(getMean(col, i))) {
avg += getMean(col, i);
count++;
}
}
return avg / (double) count;
}
else {
return 0;
}
}
/**
* sets the std deviation at the given position (if the position is valid)
* @param col the column of the std. deviation
* @param row the row of the std deviation
* @param value the value of the std deviation
*/
public void setStdDev(int col, int row, double value) {
if ( (col >= 0) && (col < getColCount())
&& (row >= 0) && (row < getRowCount()) )
m_StdDev[row][col] = value;
}
/**
* returns the std deviation at the given position, if the position is valid,
* otherwise 0
*/
public double getStdDev(int col, int row) {
if ( (col >= 0) && (col < getColCount())
&& (row >= 0) && (row < getRowCount()) )
return m_StdDev[row][col];
else
return 0;
}
/**
* sets the significance at the given position (if the position is valid)
* @param col the column of the significance
* @param row the row of the significance
* @param value the value of the significance
*/
public void setSignificance(int col, int row, int value) {
if ( (col >= 0) && (col < getColCount())
&& (row >= 0) && (row < getRowCount()) )
m_Significance[row][col] = value;
}
/**
* returns the significance at the given position, if the position is valid,
* otherwise SIGNIFICANCE_ATIE
*/
public int getSignificance(int col, int row) {
if ( (col >= 0) && (col < getColCount())
&& (row >= 0) && (row < getRowCount()) )
return m_Significance[row][col];
else
return SIGNIFICANCE_TIE;
}
/**
* counts the occurrences of the given significance type in the given
* column.
* @param col the columnn to gather the information from
* @param type the significance type, WIN/TIE/LOSS
*/
public int getSignificanceCount(int col, int type) {
int result;
int i;
result = 0;
if ( (col >= 0) && (col < getColCount()) ) {
for (i = 0; i < getRowCount(); i++) {
if (getRowHidden(i))
continue;
// no value?
if (Double.isNaN(getMean(col, i)))
continue;
if (getSignificance(col, i) == type)
result++;
}
}
return result;
}
/**
* sets the ordering of the rows, null means default
* @param order the new order of the rows
*/
public void setRowOrder(int[] order) {
int i;
// default order?
if (order == null) {
m_RowOrder = null;
}
else {
if (order.length == getRowCount()) {
m_RowOrder = new int[order.length];
for (i = 0; i < order.length; i++)
m_RowOrder[i] = order[i];
}
else {
System.err.println("setRowOrder: length does not match ("
+ order.length + " <> " + getRowCount() + ") - ignored!");
}
}
}
/**
* returns the current order of the rows, null means the default order
* @return the current order of the rows
*/
public int[] getRowOrder() {
return m_RowOrder;
}
/**
* returns the displayed index of the given row, depending on the order of
* rows, returns -1 if index out of bounds
* @param index the row to get the displayed index for
* @return the real index of the row
*/
public int getDisplayRow(int index) {
if ( (index >= 0) && (index < getRowCount()) ) {
if (getRowOrder() == null)
return index;
else
return getRowOrder()[index];
}
else {
return -1;
}
}
/**
* sets the ordering of the columns, null means default
* @param order the new order of the columns
*/
public void setColOrder(int[] order) {
int i;
// default order?
if (order == null) {
m_ColOrder = null;
}
else {
if (order.length == getColCount()) {
m_ColOrder = new int[order.length];
for (i = 0; i < order.length; i++)
m_ColOrder[i] = order[i];
}
else {
System.err.println("setColOrder: length does not match ("
+ order.length + " <> " + getColCount() + ") - ignored!");
}
}
}
/**
* returns the current order of the columns, null means the default order
* @return the current order of the columns
*/
public int[] getColOrder() {
return m_ColOrder;
}
/**
* returns the displayed index of the given col, depending on the order of
* columns, returns -1 if index out of bounds
* @param index the column to get the displayed index for
* @return the real index of the column
*/
public int getDisplayCol(int index) {
if ( (index >= 0) && (index < getColCount()) ) {
if (getColOrder() == null)
return index;
else
return getColOrder()[index];
}
else {
return -1;
}
}
/**
* returns the given number as string rounded to the given number of
* decimals. additional necessary 0's are added
* @param d the number to format
* @param prec the number of decimals after the point
* @return the formatted number
*/
protected String doubleToString(double d, int prec) {
String result;
int currentPrec;
int i;
result = Utils.doubleToString(d, prec);
// decimal point?
if (result.indexOf(".") == -1)
result += ".";
// precision so far?
currentPrec = result.length() - result.indexOf(".") - 1;
for (i = currentPrec; i < prec; i++)
result += "0";
return result;
}
/**
* trims the given string down to the given length if longer, otherwise
* leaves it unchanged. a length of "0" leaves the string always
* unchanged.
* @param s the string to trim (if too long)
* @param length the max. length (0 means infinity)
* @return the trimmed string
*/
protected String trimString(String s, int length) {
if ( (length > 0) && (s.length() > length) )
return s.substring(0, length);
else
return s;
}
/**
* pads the given string on the right until it reaches the given length, if
* longer cuts it down. if length is 0 then nothing is done.
* @param s the string to pad
* @param length the max. length of the string
* @return the padded string
*/
protected String padString(String s, int length) {
return padString(s, length, false);
}
/**
* pads the given string until it reaches the given length, if longer cuts
* it down. if length is 0 then nothing is done.
* @param s the string to pad
* @param length the max. length of the string
* @param left whether to pad left or right
* @return the padded string
*/
protected String padString(String s, int length, boolean left) {
String result;
int i;
result = s;
// pad with blanks
for (i = s.length(); i < length; i++) {
if (left)
result = " " + result;
else
result = result + " ";
}
// too long?
if ( (length > 0) && (result.length() > length) )
result = result.substring(0, length);
return result;
}
/**
* returns the length of the longest cell in the given column
* @param data the data to base the calculation on
* @param col the column to check
* @return the maximum length
*/
protected int getColSize(String[][] data, int col) {
return getColSize(data, col, false, false);
}
/**
* returns the length of the longest cell in the given column
* @param data the data to base the calculation on
* @param col the column to check
* @param skipFirst whether to skip the first row
* @param skipLast whether to skip the last row
* @return the maximum length
*/
protected int getColSize( String[][] data, int col,
boolean skipFirst, boolean skipLast ) {
int result;
int i;
result = 0;
if ( (col >= 0) && (col < data[0].length) ) {
for (i = 0; i < data.length; i++) {
// skip first?
if ( (i == 0) && (skipFirst) )
continue;
// skip last?
if ( (i == data.length - 1) && (skipLast) )
continue;
if (data[i][col].length() > result)
result = data[i][col].length();
}
}
return result;
}
/**
* removes the filter classname from the given string if it should be
* removed, otherwise leaves the string alone
* @see #getRemoveFilterName()
*/
protected String removeFilterName(String s) {
if (getRemoveFilterName())
return s.replaceAll("-weka\\.filters\\..*", "")
.replaceAll("-unsupervised\\..*", "")
.replaceAll("-supervised\\..*", "");
else
return s;
}
/**
* returns a 2-dimensional array with the prepared data. includes the column
* and row names. hidden cols/rows are already excluded. <br>
* first row: column names<br>
* last row: wins/ties/losses<br>
* first col: row names<br>
*/
protected String[][] toArray() {
int i;
int n;
int ii;
int nn;
int x;
int y;
String[][] result;
String[][] tmpResult;
int cols;
int rows;
int[] widths;
boolean valueExists;
// determine visible cols/rows
rows = getVisibleRowCount();
if (getShowAverage())
rows++;
cols = getVisibleColCount();
if (getShowStdDev())
cols = cols*3; // mean + stddev + sign.
else
cols = cols*2; // mean + stddev
result = new String[rows + 2][cols + 1];
// col names
result[0][0] = trimString("Dataset", getRowNameWidth());
x = 1;
for (ii = 0; ii < getColCount(); ii++) {
i = getDisplayCol(ii);
if (getColHidden(i))
continue;
result[0][x] = trimString(
removeFilterName(getColName(i)), getColNameWidth());
x++;
// std dev
if (getShowStdDev()) {
result[0][x] = "";
x++;
}
// sign.
result[0][x] = "";
x++;
}
// row names
y = 1;
for (ii = 0; ii < getRowCount(); ii++) {
i = getDisplayRow(ii);
if (!getRowHidden(i)) {
result[y][0] = trimString(
removeFilterName(getRowName(i)), getRowNameWidth());
y++;
}
}
// fill in mean/std dev
y = 1;
for (ii = 0; ii < getRowCount(); ii++) {
i = getDisplayRow(ii);
if (getRowHidden(i))
continue;
x = 1;
for (nn = 0; nn < getColCount(); nn++) {
n = getDisplayCol(nn);
if (getColHidden(n))
continue;
// do we have a value in the matrix?
valueExists = (!Double.isNaN(getMean(n, i)));
// mean
if (!valueExists)
result[y][x] = "";
else
result[y][x] = doubleToString(getMean(n, i), getMeanPrec());
x++;
// stddev
if (getShowStdDev()) {
if (!valueExists)
result[y][x] = "";
else if (Double.isInfinite(getStdDev(n, i)))
result[y][x] = "Inf";
else
result[y][x] = doubleToString(getStdDev(n, i), getStdDevPrec());
x++;
}
// significance
if (!valueExists) {
result[y][x] = "";
}
else {
switch (getSignificance(n, i)) {
case SIGNIFICANCE_TIE:
result[y][x] = TIE_STRING;
break;
case SIGNIFICANCE_WIN:
result[y][x] = WIN_STRING;
break;
case SIGNIFICANCE_LOSS:
result[y][x] = LOSS_STRING;
break;
}
}
x++;
}
y++;
}
// the average
if (getShowAverage()) {
y = result.length - 2;
x = 0;
result[y][0] = "Average";
x++;
for (ii = 0; ii < getColCount(); ii++) {
i = getDisplayCol(ii);
if (getColHidden(i))
continue;
// mean-average
result[y][x] = doubleToString(getAverage(i), getMeanPrec());
x++;
// std dev.
if (getShowStdDev()) {
result[y][x] = "";
x++;
}
// significance
result[y][x] = "";
x++;
}
}
// wins/ties/losses
y = result.length - 1;
x = 0;
result[y][0] = LEFT_PARENTHESES
+ WIN_STRING + "/"
+ TIE_STRING + "/"
+ LOSS_STRING
+ RIGHT_PARENTHESES;
x++;
for (ii = 0; ii < getColCount(); ii++) {
i = getDisplayCol(ii);
if (getColHidden(i))
continue;
// mean
result[y][x] = "";
x++;
// std dev.
if (getShowStdDev()) {
result[y][x] = "";
x++;
}
// significance
result[y][x] = LEFT_PARENTHESES
+ getSignificanceCount(i, SIGNIFICANCE_WIN) + "/"
+ getSignificanceCount(i, SIGNIFICANCE_TIE) + "/"
+ getSignificanceCount(i, SIGNIFICANCE_LOSS)
+ RIGHT_PARENTHESES;
x++;
}
// base column has no significance -> remove these columns
tmpResult = new String[result.length][result[0].length - 1];
x = 0;
for (i = 0; i < result[0].length; i++) {
// significance
if ( ((i == 3) && ( getShowStdDev()))
|| ((i == 2) && (!getShowStdDev())) )
continue;
for (n = 0; n < result.length; n++)
tmpResult[n][x] = result[n][i];
x++;
}
result = tmpResult;
return result;
}
/**
* returns true if the index (in the array produced by toArray(boolean))
* is the row name
*/
protected boolean isRowName(int index) {
return (index == 0);
}
/**
* returns true if the index (in the array produced by toArray(boolean))
* contains a mean
*/
protected boolean isMean(int index) {
index--; // dataset
if (index == 0) {
return true; // base column
}
else {
index--; // base column
if (index < 0)
return false;
if (getShowStdDev())
return (index % 3 == 1);
else
return (index % 2 == 0);
}
}
/**
* returns true if the row index (in the array produced by toArray(boolean))
* contains the average row
*/
protected boolean isAverage(int rowIndex) {
if (getShowAverage())
return (getVisibleRowCount() + 1 == rowIndex);
else
return false;
}
/**
* returns true if the index (in the array produced by toArray(boolean))
* contains a std deviation
*/
protected boolean isStdDev(int index) {
index--; // dataset
index--; // base column
if (getShowStdDev()) {
if (index == 0) {
return true; // stddev of base column
}
else {
index--; // stddev of base column
if (index < 0)
return false;
return (index % 3 == 1);
}
}
else
return false;
}
/**
* returns true if the index (in the array produced by toArray(boolean))
* contains a significance column
*/
protected boolean isSignificance(int index) {
index--; // dataset
index--; // base column
if (getShowStdDev()) {
index--; // stddev of base column
if (index < 0)
return false;
return (index % 3 == 2);
}
else {
if (index < 0)
return false;
return (index % 2 == 1);
}
}
/**
* returns the matrix as a string
*/
public abstract String toStringMatrix();
/**
* returns the matrix as a string
* @see #toStringMatrix()
*/
public String toString() {
return toStringMatrix();
}
/**
* removes all the header information
*/
public void clearHeader() {
m_HeaderKeys = new Vector();
m_HeaderValues = new Vector();
}
/**
* adds the key-value pair to the header
* @param key the name of the header value
* @param value the value of the header value
*/
public void addHeader(String key, String value) {
int pos;
pos = m_HeaderKeys.indexOf(key);
if (pos > -1) {
m_HeaderValues.set(pos, value);
}
else {
m_HeaderKeys.add(key);
m_HeaderValues.add(value);
}
}
/**
* returns the value associated with the given key, null if if cannot be
* found
* @param key the key to retrieve the value for
* @return the associated value
*/
public String getHeader(String key) {
int pos;
pos = m_HeaderKeys.indexOf(key);
if (pos == 0)
return null;
else
return (String) m_HeaderKeys.get(pos);
}
/**
* returns an enumeration of the header keys
* @return all stored keys
*/
public Enumeration headerKeys() {
return m_HeaderKeys.elements();
}
/**
* returns the header of the matrix as a string
* @see #m_HeaderKeys
* @see #m_HeaderValues
*/
public abstract String toStringHeader();
/**
* returns returns a key for all the col names, for better readability if
* the names got cut off
*/
public abstract String toStringKey();
/**
* clears the current summary data
*/
public void clearSummary() {
m_NonSigWins = null;
m_Wins = null;
}
/**
* sets the non-significant and significant wins of the resultsets
* @param nonSigWins the non-significant wins
* @param wins the significant wins
*/
public void setSummary(int[][] nonSigWins, int[][] wins) {
int i;
int n;
m_NonSigWins = new int[nonSigWins.length][nonSigWins[0].length];
m_Wins = new int[wins.length][wins[0].length];
for (i = 0; i < m_NonSigWins.length; i++) {
for (n = 0; n < m_NonSigWins[i].length; n++) {
m_NonSigWins[i][n] = nonSigWins[i][n];
m_Wins[i][n] = wins[i][n];
}
}
}
/**
* returns the character representation of the given column
*/
protected String getSummaryTitle(int col) {
return "" + (char) ((int) 'a' + col % 26);
}
/**
* returns the summary as string
*/
public abstract String toStringSummary();
/**
* clears the currently stored ranking data
*/
public void clearRanking() {
m_RankingWins = null;
m_RankingLosses = null;
m_RankingDiff = null;
}
/**
* sets the ranking data based on the wins
* @param wins the wins
*/
public void setRanking(int[][] wins) {
int i;
int j;
m_RankingWins = new int[wins.length];
m_RankingLosses = new int[wins.length];
m_RankingDiff = new int[wins.length];
for (i = 0; i < wins.length; i++) {
for (j = 0; j < wins[i].length; j++) {
m_RankingWins[j] += wins[i][j];
m_RankingDiff[j] += wins[i][j];
m_RankingLosses[i] += wins[i][j];
m_RankingDiff[i] -= wins[i][j];
}
}
}
/**
* returns the ranking in a string representation
*/
public abstract String toStringRanking();
}