/**
* Computes average class values for each attribute and value
*
* @throws Exception
*/
private void computeAverageClassValues() throws Exception {
double totalCounts, sum;
M5Instance instance;
double[] counts;
double[][] avgClassValues = new double[getInputFormat().numAttributes()][0];
m_Indices = new int[getInputFormat().numAttributes()][0];
for (int j = 0; j < getInputFormat().numAttributes(); j++) {
M5Attribute att = getInputFormat().attribute(j);
if (att.isNominal()) {
avgClassValues[j] = new double[att.numValues()];
counts = new double[att.numValues()];
for (int i = 0; i < getInputFormat().numInstances(); i++) {
instance = getInputFormat().instance(i);
if (!instance.classIsMissing() && (!instance.isMissing(j))) {
counts[(int) instance.value(j)] += instance.weight();
avgClassValues[j][(int) instance.value(j)] += instance.weight() * instance.classValue();
}
}
sum = M5StaticUtils.sum(avgClassValues[j]);
totalCounts = M5StaticUtils.sum(counts);
if (M5StaticUtils.gr(totalCounts, 0)) {
for (int k = 0; k < att.numValues(); k++) {
if (M5StaticUtils.gr(counts[k], 0)) {
avgClassValues[j][k] /= (double) counts[k];
} else {
avgClassValues[j][k] = sum / (double) totalCounts;
}
}
}
m_Indices[j] = M5StaticUtils.sort(avgClassValues[j]);
}
}
}
/**
* Converts carriage returns and new lines in a string into \r and \n. Backquotes the following
* characters: ` " \ \t and %
*
* @param string the string
* @return the converted string
*/
public static String backQuoteChars(String string) {
int index;
StringBuffer newStringBuffer;
// replace each of the following characters with the backquoted version
char charsFind[] = {'\\', '\'', '\t', '"', '%'};
String charsReplace[] = {"\\\\", "\\'", "\\t", "\\\"", "\\%"};
for (int i = 0; i < charsFind.length; i++) {
if (string.indexOf(charsFind[i]) != -1) {
newStringBuffer = new StringBuffer();
while ((index = string.indexOf(charsFind[i])) != -1) {
if (index > 0) {
newStringBuffer.append(string.substring(0, index));
}
newStringBuffer.append(charsReplace[i]);
if ((index + 1) < string.length()) {
string = string.substring(index + 1);
} else {
string = "";
}
}
newStringBuffer.append(string);
string = newStringBuffer.toString();
}
}
return M5StaticUtils.convertNewLines(string);
}
/**
* Returns c*log2(c) for a given integer value c.
*
* @param c an integer value
* @returns c*log2(c) (but is careful to return 0 if c is 0)
*/
public static double xlogx(int c) {
if (c == 0) {
return 0.0;
}
return c * M5StaticUtils.log2((double) c);
}
/**
* Parses the options for this object. Valid options are:
*
* <p>-N <br>
* If binary attributes are to be coded as nominal ones.
*
* <p>
*
* @param options the list of options as an array of strings
* @exception Exception if an option is not supported
*/
public void setOptions(String[] options) throws Exception {
setBinaryAttributesNominal(M5StaticUtils.getFlag('N', options));
if (getInputFormat() != null) {
setInputFormat(getInputFormat());
}
}
/**
* Sorts a given array of doubles in ascending order and returns an array of integers with the
* positions of the elements of the original array in the sorted array. The sort is stable (Equal
* elements remain in their original order.) Occurrences of Double.NaN are treated as
* Double.MAX_VALUE
*
* @param array this array is not changed by the method!
* @return an array of integers with the positions in the sorted array.
*/
public static int[] stableSort(double[] array) {
int[] index = new int[array.length];
int[] newIndex = new int[array.length];
int[] helpIndex;
int numEqual;
array = (double[]) array.clone();
for (int i = 0; i < index.length; i++) {
index[i] = i;
if (Double.isNaN(array[i])) {
array[i] = Double.MAX_VALUE;
}
}
quickSort(array, index, 0, array.length - 1);
// Make sort stable
int i = 0;
while (i < index.length) {
numEqual = 1;
for (int j = i + 1;
((j < index.length) && M5StaticUtils.eq(array[index[i]], array[index[j]]));
j++) {
numEqual++;
}
if (numEqual > 1) {
helpIndex = new int[numEqual];
for (int j = 0; j < numEqual; j++) {
helpIndex[j] = i + j;
}
quickSort(index, helpIndex, 0, numEqual - 1);
for (int j = 0; j < numEqual; j++) {
newIndex[i + j] = index[helpIndex[j]];
}
i += numEqual;
} else {
newIndex[i] = index[i];
i++;
}
}
return newIndex;
}
/**
* Creates a new instance of an object given it's class name and (optional) arguments to pass to
* it's setOptions method. If the object implements OptionHandler and the options parameter is
* non-null, the object will have it's options set. Example use:
*
* <p><code> <pre>
* String classifierName = M5StaticUtils.getOption('W', options);
* Classifier c = (Classifier)M5StaticUtils.forName(Classifier.class,
* classifierName,
* options);
* setClassifier(c);
* </pre></code>
*
* @param classType the class that the instantiated object should be assignable to -- an exception
* is thrown if this is not the case
* @param className the fully qualified class name of the object
* @param options an array of options suitable for passing to setOptions. May be null. Any options
* accepted by the object will be removed from the array.
* @return the newly created object, ready for use.
* @exception Exception if the class name is invalid, or if the class is not assignable to the
* desired class type, or the options supplied are not acceptable to the object
*/
public static Object forName(Class classType, String className, String[] options)
throws Exception {
Class c = null;
try {
c = Class.forName(className);
} catch (Exception ex) {
throw new Exception("Can't find class called: " + className);
}
if (!classType.isAssignableFrom(c)) {
throw new Exception(classType.getName() + " is not assignable from " + className);
}
Object o = c.newInstance();
if ((o instanceof M5) && (options != null)) {
((M5) o).setOptions(options);
M5StaticUtils.checkForRemainingOptions(options);
}
return o;
}
/** Display a representation of this estimator */
public String toString() {
String result =
m_NumValues
+ " Normal Kernels. \nStandardDev = "
+ M5StaticUtils.doubleToString(m_StandardDev, 6, 4)
+ " Precision = "
+ m_Precision;
if (m_NumValues == 0) {
result += " \nMean = 0";
} else {
result += " \nMeans =";
for (int i = 0; i < m_NumValues; i++) {
result += " " + m_Values[i];
}
if (!m_AllWeightsOne) {
result += "\nWeights = ";
for (int i = 0; i < m_NumValues; i++) {
result += " " + m_Weights[i];
}
}
}
return result + "\n";
}
/**
* Main method for testing this class.
*
* @param ops some dummy options
*/
public static void main(String[] ops) {
double[] doubles = {4.5, 6.7, Double.NaN, 3.4, 4.8, 1.2, 3.4};
int[] ints = {12, 6, 2, 18, 16, 6, 7, 5};
try {
// Option handling
System.out.println("First option split up:");
if (ops.length > 0) {
String[] firstOptionSplitUp = M5StaticUtils.splitOptions(ops[0]);
for (int i = 0; i < firstOptionSplitUp.length; i++) {
System.out.println(firstOptionSplitUp[i]);
}
}
System.out.println("Partitioned options: ");
String[] partitionedOptions = M5StaticUtils.partitionOptions(ops);
for (int i = 0; i < partitionedOptions.length; i++) {
System.out.println(partitionedOptions[i]);
}
System.out.println("Get flag -f: " + M5StaticUtils.getFlag('f', ops));
System.out.println("Get option -o: " + M5StaticUtils.getOption('o', ops));
System.out.println("Checking for remaining options... ");
M5StaticUtils.checkForRemainingOptions(ops);
// Statistics
System.out.println("Original array (doubles): ");
for (int i = 0; i < doubles.length; i++) {
System.out.print(doubles[i] + " ");
}
System.out.println();
System.out.println("Original array (ints): ");
for (int i = 0; i < ints.length; i++) {
System.out.print(ints[i] + " ");
}
System.out.println();
System.out.println(
"Correlation: " + M5StaticUtils.correlation(doubles, doubles, doubles.length));
System.out.println("Mean: " + M5StaticUtils.mean(doubles));
System.out.println("Variance: " + M5StaticUtils.variance(doubles));
System.out.println("Sum (doubles): " + M5StaticUtils.sum(doubles));
System.out.println("Sum (ints): " + M5StaticUtils.sum(ints));
System.out.println("Max index (doubles): " + M5StaticUtils.maxIndex(doubles));
System.out.println("Max index (ints): " + M5StaticUtils.maxIndex(ints));
System.out.println("Min index (doubles): " + M5StaticUtils.minIndex(doubles));
System.out.println("Min index (ints): " + M5StaticUtils.minIndex(ints));
// Sorting and normalizing
System.out.println("Sorted array (doubles): ");
int[] sorted = M5StaticUtils.sort(doubles);
for (int i = 0; i < doubles.length; i++) {
System.out.print(doubles[sorted[i]] + " ");
}
System.out.println();
System.out.println("Normalized array (doubles): ");
M5StaticUtils.normalize(doubles);
for (int i = 0; i < doubles.length; i++) {
System.out.print(doubles[i] + " ");
}
System.out.println();
System.out.println("Normalized again (doubles): ");
M5StaticUtils.normalize(doubles, M5StaticUtils.sum(doubles));
for (int i = 0; i < doubles.length; i++) {
System.out.print(doubles[i] + " ");
}
System.out.println();
// Pretty-printing
System.out.println("-4.58: " + M5StaticUtils.doubleToString(-4.57826535, 2));
System.out.println("-6.78: " + M5StaticUtils.doubleToString(-6.78214234, 6, 2));
// Comparisons
System.out.println("5.70001 == 5.7 ? " + M5StaticUtils.eq(5.70001, 5.7));
System.out.println("5.70001 > 5.7 ? " + M5StaticUtils.gr(5.70001, 5.7));
System.out.println("5.70001 >= 5.7 ? " + M5StaticUtils.grOrEq(5.70001, 5.7));
System.out.println("5.7 < 5.70001 ? " + M5StaticUtils.sm(5.7, 5.70001));
System.out.println("5.7 <= 5.70001 ? " + M5StaticUtils.smOrEq(5.7, 5.70001));
// Math
System.out.println("Info (ints): " + M5StaticUtils.info(ints));
System.out.println("log2(4.6): " + M5StaticUtils.log2(4.6));
System.out.println("5 * log(5): " + M5StaticUtils.xlogx(5));
System.out.println("5.5 rounded: " + M5StaticUtils.round(5.5));
System.out.println(
"5.55555 rounded to 2 decimal places: " + M5StaticUtils.roundDouble(5.55555, 2));
} catch (Exception e) {
e.printStackTrace();
}
}
/**
* Method for testing filters ability to process multiple batches.
*
* @param options should contain the following arguments:<br>
* -i (first) input file <br>
* -o (first) output file <br>
* -r (second) input file <br>
* -s (second) output file <br>
* -c class_index <br>
* or -h for help on options
* @exception Exception if something goes wrong or the user requests help on command options
*/
public static void batchFilterFile(NominalToBinaryFilter filter, String[] options)
throws Exception {
M5Instances firstData = null;
M5Instances secondData = null;
Reader firstInput = null;
Reader secondInput = null;
PrintWriter firstOutput = null;
PrintWriter secondOutput = null;
boolean helpRequest;
try {
helpRequest = M5StaticUtils.getFlag('h', options);
String fileName = M5StaticUtils.getOption('i', options);
if (fileName.length() != 0) {
firstInput = new BufferedReader(new FileReader(fileName));
} else {
throw new Exception("No first input file given.\n");
}
fileName = M5StaticUtils.getOption('r', options);
if (fileName.length() != 0) {
secondInput = new BufferedReader(new FileReader(fileName));
} else {
throw new Exception("No second input file given.\n");
}
fileName = M5StaticUtils.getOption('o', options);
if (fileName.length() != 0) {
firstOutput = new PrintWriter(new FileOutputStream(fileName));
} else {
firstOutput = new PrintWriter(System.out);
}
fileName = M5StaticUtils.getOption('s', options);
if (fileName.length() != 0) {
secondOutput = new PrintWriter(new FileOutputStream(fileName));
} else {
secondOutput = new PrintWriter(System.out);
}
String classIndex = M5StaticUtils.getOption('c', options);
if (filter instanceof NominalToBinaryFilter) {
((NominalToBinaryFilter) filter).setOptions(options);
}
M5StaticUtils.checkForRemainingOptions(options);
if (helpRequest) {
throw new Exception("Help requested.\n");
}
firstData = new M5Instances(firstInput, 1);
secondData = new M5Instances(secondInput, 1);
if (!secondData.equalHeaders(firstData)) {
throw new Exception("Input file formats differ.\n");
}
if (classIndex.length() != 0) {
if (classIndex.equals("first")) {
firstData.setClassIndex(0);
secondData.setClassIndex(0);
} else if (classIndex.equals("last")) {
firstData.setClassIndex(firstData.numAttributes() - 1);
secondData.setClassIndex(secondData.numAttributes() - 1);
} else {
firstData.setClassIndex(Integer.parseInt(classIndex) - 1);
secondData.setClassIndex(Integer.parseInt(classIndex) - 1);
}
}
} catch (Exception ex) {
String filterOptions = "";
// Output the error and also the valid options
if (filter instanceof NominalToBinaryFilter) {
filterOptions += "\nFilter options:\n\n";
Enumeration enume = ((NominalToBinaryFilter) filter).listOptions();
while (enume.hasMoreElements()) {
Information option = (Information) enume.nextElement();
filterOptions += option.synopsis() + '\n' + option.description() + "\n";
}
}
String genericOptions =
"\nGeneral options:\n\n"
+ "-h\n"
+ "\tGet help on available options.\n"
+ "-i <filename>\n"
+ "\tThe file containing first input instances.\n"
+ "-o <filename>\n"
+ "\tThe file first output instances will be written to.\n"
+ "-r <filename>\n"
+ "\tThe file containing second input instances.\n"
+ "-s <filename>\n"
+ "\tThe file second output instances will be written to.\n"
+ "-c <class index>\n"
+ "\tThe number of the attribute to use as the class.\n"
+ "\t\"first\" and \"last\" are also valid entries.\n"
+ "\tIf not supplied then no class is assigned.\n";
throw new Exception('\n' + ex.getMessage() + filterOptions + genericOptions);
}
boolean printedHeader = false;
if (filter.setInputFormat(firstData)) {
firstOutput.println(filter.getOutputFormat().toString());
printedHeader = true;
}
// Pass all the instances to the filter
while (firstData.readInstance(firstInput)) {
if (filter.input(firstData.instance(0))) {
if (!printedHeader) {
throw new Error("Filter didn't return true from setInputFormat() " + "earlier!");
}
firstOutput.println(filter.output().toString());
}
firstData.delete(0);
}
// Say that input has finished, and print any pending output instances
if (filter.batchFinished()) {
if (!printedHeader) {
firstOutput.println(filter.getOutputFormat().toString());
}
while (filter.numPendingOutput() > 0) {
firstOutput.println(filter.output().toString());
}
}
if (firstOutput != null) {
firstOutput.close();
}
printedHeader = false;
if (filter.isOutputFormatDefined()) {
secondOutput.println(filter.getOutputFormat().toString());
printedHeader = true;
}
// Pass all the second instances to the filter
while (secondData.readInstance(secondInput)) {
if (filter.input(secondData.instance(0))) {
if (!printedHeader) {
throw new Error("Filter didn't return true from" + " isOutputFormatDefined() earlier!");
}
secondOutput.println(filter.output().toString());
}
secondData.delete(0);
}
// Say that input has finished, and print any pending output instances
if (filter.batchFinished()) {
if (!printedHeader) {
secondOutput.println(filter.getOutputFormat().toString());
}
while (filter.numPendingOutput() > 0) {
secondOutput.println(filter.output().toString());
}
}
if (secondOutput != null) {
secondOutput.close();
}
}
/**
* Method for testing filters.
*
* @param options should contain the following arguments: <br>
* -i input_file <br>
* -o output_file <br>
* -c class_index <br>
* or -h for help on options
* @exception Exception if something goes wrong or the user requests help on command options
*/
public static void filterFile(NominalToBinaryFilter filter, String[] options) throws Exception {
boolean debug = false;
M5Instances data = null;
Reader input = null;
PrintWriter output = null;
boolean helpRequest;
try {
helpRequest = M5StaticUtils.getFlag('h', options);
if (M5StaticUtils.getFlag('d', options)) {
debug = true;
}
String infileName = M5StaticUtils.getOption('i', options);
String outfileName = M5StaticUtils.getOption('o', options);
String classIndex = M5StaticUtils.getOption('c', options);
if (filter instanceof NominalToBinaryFilter) {
((NominalToBinaryFilter) filter).setOptions(options);
}
M5StaticUtils.checkForRemainingOptions(options);
if (helpRequest) {
throw new Exception("Help requested.\n");
}
if (infileName.length() != 0) {
input = new BufferedReader(new FileReader(infileName));
} else {
input = new BufferedReader(new InputStreamReader(System.in));
}
if (outfileName.length() != 0) {
output = new PrintWriter(new FileOutputStream(outfileName));
} else {
output = new PrintWriter(System.out);
}
data = new M5Instances(input, 1);
if (classIndex.length() != 0) {
if (classIndex.equals("first")) {
data.setClassIndex(0);
} else if (classIndex.equals("last")) {
data.setClassIndex(data.numAttributes() - 1);
} else {
data.setClassIndex(Integer.parseInt(classIndex) - 1);
}
}
} catch (Exception ex) {
String filterOptions = "";
// Output the error and also the valid options
if (filter instanceof NominalToBinaryFilter) {
filterOptions += "\nFilter options:\n\n";
Enumeration enuma = ((NominalToBinaryFilter) filter).listOptions();
while (enuma.hasMoreElements()) {
Information option = (Information) enuma.nextElement();
filterOptions += option.synopsis() + '\n' + option.description() + "\n";
}
}
String genericOptions =
"\nGeneral options:\n\n"
+ "-h\n"
+ "\tGet help on available options.\n"
+ "\t(use -b -h for help on batch mode.)\n"
+ "-i <file>\n"
+ "\tThe name of the file containing input instances.\n"
+ "\tIf not supplied then instances will be read from stdin.\n"
+ "-o <file>\n"
+ "\tThe name of the file output instances will be written to.\n"
+ "\tIf not supplied then instances will be written to stdout.\n"
+ "-c <class index>\n"
+ "\tThe number of the attribute to use as the class.\n"
+ "\t\"first\" and \"last\" are also valid entries.\n"
+ "\tIf not supplied then no class is assigned.\n";
throw new Exception('\n' + ex.getMessage() + filterOptions + genericOptions);
}
if (debug) {
System.err.println("Setting input format");
}
boolean printedHeader = false;
if (filter.setInputFormat(data)) {
if (debug) {
System.err.println("Getting output format");
}
output.println(filter.getOutputFormat().toString());
printedHeader = true;
}
// Pass all the instances to the filter
while (data.readInstance(input)) {
if (debug) {
System.err.println("Input instance to filter");
}
if (filter.input(data.instance(0))) {
if (debug) {
System.err.println("Filter said collect immediately");
}
if (!printedHeader) {
throw new Error("Filter didn't return true from setInputFormat() " + "earlier!");
}
if (debug) {
System.err.println("Getting output instance");
}
output.println(filter.output().toString());
}
data.delete(0);
}
// Say that input has finished, and print any pending output instances
if (debug) {
System.err.println("Setting end of batch");
}
if (filter.batchFinished()) {
if (debug) {
System.err.println("Filter said collect output");
}
if (!printedHeader) {
if (debug) {
System.err.println("Getting output format");
}
output.println(filter.getOutputFormat().toString());
}
if (debug) {
System.err.println("Getting output instance");
}
while (filter.numPendingOutput() > 0) {
output.println(filter.output().toString());
if (debug) {
System.err.println("Getting output instance");
}
}
}
if (debug) {
System.err.println("Done");
}
if (output != null) {
output.close();
}
}
