@Override
protected MDInteger getSampledSize(ExampleSetMetaData emd) throws UndefinedParameterError {
switch (getParameterAsInt(PARAMETER_SAMPLE)) {
case SAMPLE_ABSOLUTE:
int absoluteNumber = getParameterAsInt(PARAMETER_SAMPLE_SIZE);
if (emd.getNumberOfExamples().isAtLeast(absoluteNumber) == MetaDataInfo.NO)
getExampleSetInputPort()
.addError(
new SimpleMetaDataError(
Severity.ERROR,
getExampleSetInputPort(),
Collections.singletonList(
new ParameterSettingQuickFix(
this,
PARAMETER_SAMPLE_SIZE,
emd.getNumberOfExamples().getValue().toString())),
"need_more_examples",
absoluteNumber + ""));
return new MDInteger(absoluteNumber);
case SAMPLE_RELATIVE:
MDInteger number = emd.getNumberOfExamples();
number.multiply(getParameterAsDouble(PARAMETER_SAMPLE_RATIO));
return number;
default:
return new MDInteger();
}
}
public ProcessLog2ExampleSet(OperatorDescription description) {
super(description);
ExampleSetMetaData newEMD = new ExampleSetMetaData();
newEMD.attributesAreSuperset();
newEMD.setNumberOfExamples(0);
newEMD.getNumberOfExamples().increaseByUnknownAmount();
getTransformer().addRule(new GenerateNewMDRule(exampleSetOutput, newEMD));
dummyPorts.start();
getTransformer().addRule(dummyPorts.makePassThroughRule());
}
@Override
protected MDInteger getSampledSize(ExampleSetMetaData emd) throws UndefinedParameterError {
switch (getParameterAsInt(PARAMETER_SAMPLE)) {
case SAMPLE_ABSOLUTE:
return new MDInteger(getParameterAsInt(PARAMETER_SAMPLE_SIZE));
case SAMPLE_RELATIVE:
MDInteger number = emd.getNumberOfExamples();
number.multiply(getParameterAsDouble(PARAMETER_SAMPLE_RATIO));
return number;
default:
return new MDInteger();
}
}
@Override
protected MetaData modifyMetaData(ExampleSetMetaData metaData) throws UndefinedParameterError {
ExampleSetMetaData resultMD = metaData.clone();
resultMD.clear();
// add group by attributes
if (isParameterSet(PARAMETER_GROUP_BY_ATTRIBUTES)
&& !getParameterAsString(PARAMETER_GROUP_BY_ATTRIBUTES).isEmpty()) {
String attributeRegex = getParameterAsString(PARAMETER_GROUP_BY_ATTRIBUTES);
Pattern pattern = Pattern.compile(attributeRegex);
for (AttributeMetaData amd : metaData.getAllAttributes()) {
if (pattern.matcher(amd.getName()).matches()) {
if (amd.isNumerical()
&& getCompatibilityLevel().isAtMost(VERSION_5_1_6)) { // converting type to mimic
// NumericalToPolynomial used below
amd.setType(Ontology.NOMINAL);
amd.setValueSet(Collections.<String>emptySet(), SetRelation.SUPERSET);
}
resultMD.addAttribute(amd);
}
}
resultMD.getNumberOfExamples().reduceByUnknownAmount();
}
if (resultMD.getAllAttributes().isEmpty() && getCompatibilityLevel().isAtMost(VERSION_5_1_6)) {
AttributeMetaData allGroup = new AttributeMetaData(GENERIC_GROUP_NAME, Ontology.NOMINAL);
Set<String> values = new TreeSet<String>();
values.add(GENERIC_ALL_NAME);
allGroup.setValueSet(values, SetRelation.EQUAL);
resultMD.addAttribute(allGroup);
resultMD.setNumberOfExamples(new MDInteger(1));
}
// add aggregated attributes of default aggregation: They will apply only to those attribute not
// mentioned explicitly
List<String[]> parameterList = this.getParameterList(PARAMETER_AGGREGATION_ATTRIBUTES);
HashSet<String> explicitDefinedAttributes = new HashSet<String>();
for (String[] function : parameterList) {
explicitDefinedAttributes.add(function[0]);
}
if (getParameterAsBoolean(PARAMETER_USE_DEFAULT_AGGREGATION)) {
String defaultFunction = getParameterAsString(PARAMETER_DEFAULT_AGGREGATION_FUNCTION);
ExampleSetMetaData metaDataSubset = attributeSelector.getMetaDataSubset(metaData, false);
for (AttributeMetaData amd : metaDataSubset.getAllAttributes()) {
if (!explicitDefinedAttributes.contains(amd.getName())) {
AttributeMetaData newAMD =
AggregationFunction.getAttributeMetaData(
defaultFunction, amd, getExampleSetInputPort());
if (newAMD != null) resultMD.addAttribute(newAMD);
}
}
}
// add explicitly defined attributes of list
for (String[] function : parameterList) {
AttributeMetaData amd = metaData.getAttributeByName(function[0]);
if (amd != null) {
AttributeMetaData newMD =
AggregationFunction.getAttributeMetaData(function[1], amd, getExampleSetInputPort());
if (newMD != null) resultMD.addAttribute(newMD);
} else {
// in this case we should register a warning, but continue anyway in cases we don't have the
// correct set available
getExampleSetInputPort()
.addError(
new SimpleMetaDataError(
Severity.WARNING,
getExampleSetInputPort(),
"aggregation.attribute_unknown",
function[0]));
AttributeMetaData newAMD =
AggregationFunction.getAttributeMetaData(
function[1],
new AttributeMetaData(function[0], Ontology.ATTRIBUTE_VALUE),
getExampleSetInputPort());
if (newAMD != null) resultMD.addAttribute(newAMD);
}
}
return resultMD;
}
@Override
protected MetaData modifyMetaData(ExampleSetMetaData metaData) throws UndefinedParameterError {
metaData.getNumberOfExamples().reduceByUnknownAmount();
// TODO: Could instead take a look at the values of ids if nominal.
return metaData;
}