public static Double getAffinity(Object object, OutputField outputField) {
HasAffinity hasAffinity = TypeUtil.cast(HasAffinity.class, object);
int rank = outputField.getRank();
if (rank <= 0) {
throw new InvalidFeatureException(outputField);
}
if (rank > 1) {
HasAffinityRanking hasAffinityRanking = TypeUtil.cast(HasAffinityRanking.class, object);
OutputField.RankOrder rankOrder = outputField.getRankOrder();
switch (rankOrder) {
case DESCENDING:
break;
default:
throw new UnsupportedFeatureException(outputField, rankOrder);
}
return getElement(hasAffinityRanking.getAffinityRanking(), rank);
}
String value = getCategoryValue(object, outputField);
return hasAffinity.getAffinity(value);
}
public static Double getCategoricalResidual(Object object, FieldValue expectedObject) {
HasProbability hasProbability = TypeUtil.cast(HasProbability.class, object);
String value = TypeUtil.format(getPredictedValue(object));
String expectedValue = TypeUtil.format(FieldValueUtil.getValue(expectedObject));
boolean equals = TypeUtil.equals(DataType.STRING, value, expectedValue);
return Double.valueOf((equals ? 1d : 0d) - hasProbability.getProbability(value));
}
public static TargetValue getTargetValue(Target target, Object value) {
DataType dataType = TypeUtil.getDataType(value);
List<TargetValue> targetValues = target.getTargetValues();
for (TargetValue targetValue : targetValues) {
if (TypeUtil.equals(
dataType, value, TypeUtil.parseOrCast(dataType, targetValue.getValue()))) {
return targetValue;
}
}
return null;
}
private static Double getProbability(Object object, OutputField outputField) {
HasProbability hasProbability = TypeUtil.cast(HasProbability.class, object);
String value = getCategoryValue(object, outputField);
return hasProbability.getProbability(value);
}
@Override
public FieldValue evaluate(List<FieldValue> values) {
if (values.size() != 2) {
throw new EvaluationException();
}
FieldValue left = values.get(0);
FieldValue right = values.get(1);
// "If one of the input fields of a simple arithmetic function is a missing value, the result
// evaluates to missing value"
if (left == null || right == null) {
return null;
}
DataType dataType = TypeUtil.getResultDataType(left.getDataType(), right.getDataType());
Number result;
try {
result = evaluate(left.asNumber(), right.asNumber());
} catch (ArithmeticException ae) {
throw new InvalidResultException(null);
}
return FieldValueUtil.create(cast(dataType, result));
}
@Override
public FieldValue evaluate(List<FieldValue> values) {
StorelessUnivariateStatistic statistic = createStatistic();
DataType dataType = null;
for (FieldValue value : values) {
// "Missing values in the input to an aggregate function are simply ignored"
if (value == null) {
continue;
}
statistic.increment((value.asNumber()).doubleValue());
if (dataType != null) {
dataType = TypeUtil.getResultDataType(dataType, value.getDataType());
} else {
dataType = value.getDataType();
}
}
if (statistic.getN() == 0) {
throw new MissingResultException(null);
}
Object result = cast(getResultType(dataType), statistic.getResult());
return FieldValueUtil.create(result);
}
public static String getReasonCode(Object object, OutputField outputField) {
HasReasonCodeRanking hasReasonCodeRanking = TypeUtil.cast(HasReasonCodeRanking.class, object);
int rank = outputField.getRank();
if (rank <= 0) {
throw new InvalidFeatureException(outputField);
}
return getElement(hasReasonCodeRanking.getReasonCodeRanking(), rank);
}
public static Object getRuleValue(Object object, OutputField outputField) {
HasRuleValues hasRuleValues = TypeUtil.cast(HasRuleValues.class, object);
List<AssociationRule> associationRules = getRuleValues(hasRuleValues, outputField);
String isMultiValued = outputField.getIsMultiValued();
// Return a single result
if (("0").equals(isMultiValued)) {
int rank = outputField.getRank();
if (rank <= 0) {
throw new InvalidFeatureException(outputField);
}
AssociationRule associationRule = getElement(associationRules, rank);
if (associationRule != null) {
return getRuleFeature(hasRuleValues, associationRule, outputField);
}
return null;
} else
// Return multiple results
if (("1").equals(isMultiValued)) {
int size;
int rank = outputField.getRank();
if (rank < 0) {
throw new InvalidFeatureException(outputField);
} else
// "A zero value indicates that all output values are to be returned"
if (rank == 0) {
size = associationRules.size();
} else
// "A positive value indicates the number of output values to be returned"
{
size = Math.min(rank, associationRules.size());
}
associationRules = associationRules.subList(0, size);
List<Object> result = new ArrayList<>(associationRules.size());
for (AssociationRule associationRule : associationRules) {
result.add(getRuleFeature(hasRuleValues, associationRule, outputField));
}
return result;
} else {
throw new InvalidFeatureException(outputField);
}
}
private static String getCategoryValue(Object object, OutputField outputField) {
String value = outputField.getValue();
// "If the value attribute is not specified, then the predicted categorical value should be
// returned as a result"
if (value == null) {
return TypeUtil.format(getPredictedValue(object));
}
return value;
}
static Object evaluateRegressionInternal(
FieldName name, Object value, ModelEvaluationContext context) {
ModelEvaluator<?> evaluator = context.getModelEvaluator();
if (Objects.equals(Evaluator.DEFAULT_TARGET, name)) {
DataField dataField = evaluator.getDataField();
if (value != null) {
value = TypeUtil.cast(dataField.getDataType(), value);
}
} else {
Target target = evaluator.getTarget(name);
if (target != null) {
if (value == null) {
value = getDefaultValue(target);
} // End if
if (value != null) {
value = processValue(target, (Double) value);
}
}
DataField dataField = evaluator.getDataField(name);
if (dataField == null) {
throw new MissingFieldException(name);
} // End if
if (value != null) {
value = TypeUtil.cast(dataField.getDataType(), value);
}
MiningField miningField = evaluator.getMiningField(name);
context.declare(
name, FieldValueUtil.createTargetValue(dataField, miningField, target, value));
}
return value;
}
private static Object getPredictedDisplayValue(
Object object, DataField dataField, Target target) {
if (object instanceof HasDisplayValue) {
HasDisplayValue hasDisplayValue = TypeUtil.cast(HasDisplayValue.class, object);
return hasDisplayValue.getDisplayValue();
}
object = getPredictedValue(object);
if (target != null) {
TargetValue targetValue = TargetUtil.getTargetValue(target, object);
if (targetValue != null) {
String displayValue = targetValue.getDisplayValue();
if (displayValue != null) {
return displayValue;
}
}
}
OpType opType = dataField.getOpType();
switch (opType) {
case CONTINUOUS:
break;
case CATEGORICAL:
case ORDINAL:
{
Value value = FieldValueUtil.getValidValue(dataField, object);
if (value != null) {
String displayValue = value.getDisplayValue();
if (displayValue != null) {
return displayValue;
}
}
}
break;
default:
throw new UnsupportedFeatureException(dataField, opType);
}
// "If the display value is not specified explicitly, then the raw predicted value is used by
// default"
return object;
}
private static String getEntityId(Object object, OutputField outputField) {
HasEntityId hasEntityId = TypeUtil.cast(HasEntityId.class, object);
int rank = outputField.getRank();
if (rank <= 0) {
throw new InvalidFeatureException(outputField);
}
if (rank > 1) {
HasEntityIdRanking hasEntityIdRanking = TypeUtil.cast(HasEntityIdRanking.class, object);
OutputField.RankOrder rankOrder = outputField.getRankOrder();
switch (rankOrder) {
case DESCENDING:
break;
default:
throw new UnsupportedFeatureException(outputField, rankOrder);
}
return getElement(hasEntityIdRanking.getEntityIdRanking(), rank);
}
return hasEntityId.getEntityId();
}
private Map<FieldName, ?> evaluateRegression(ModelEvaluationContext context) {
Trail trail = new Trail();
Node node = evaluateTree(trail, context);
if (node == null) {
return TargetUtil.evaluateRegressionDefault(context);
}
Double score = (Double) TypeUtil.parseOrCast(DataType.DOUBLE, node.getScore());
TargetField targetField = getTargetField();
NodeScore nodeScore =
createNodeScore(node, TargetUtil.evaluateRegressionInternal(targetField, score, context));
return Collections.singletonMap(targetField.getName(), nodeScore);
}
private static Double evaluate(Collection<?> values, int quantile) {
List<Double> doubleValues = new ArrayList<>();
for (Object value : values) {
Double doubleValue = (Double) TypeUtil.parseOrCast(DataType.DOUBLE, value);
doubleValues.add(doubleValue);
}
double[] data = Doubles.toArray(doubleValues);
// The data must be (at least partially) ordered
Arrays.sort(data);
Percentile percentile = new Percentile();
percentile.setData(data);
return percentile.evaluate(quantile);
}
public static Object getRuleValue(
Object object, OutputField outputField, OutputField.RuleFeature ruleFeature) {
HasRuleValues hasRuleValues = TypeUtil.cast(HasRuleValues.class, object);
List<AssociationRule> associationRules = getRuleValues(hasRuleValues, outputField);
String isMultiValued = outputField.getIsMultiValued();
if (!("0").equals(isMultiValued)) {
throw new UnsupportedFeatureException(outputField);
}
int rank = outputField.getRank();
if (rank <= 0) {
throw new InvalidFeatureException(outputField);
}
AssociationRule associationRule = getElement(associationRules, rank);
if (associationRule != null) {
return getRuleFeature(hasRuleValues, associationRule, outputField, ruleFeature);
}
return null;
}
public static Double getEntityAffinity(Object object) {
HasEntityAffinity hasEntityAffinity = TypeUtil.cast(HasEntityAffinity.class, object);
return hasEntityAffinity.getEntityAffinity();
}
/**
* Evaluates the {@link Output} element.
*
* @param predictions A map of {@link Evaluator#getTargetFields() target field} values.
* @return A map of {@link Evaluator#getTargetFields() target field} values together with {@link
* Evaluator#getOutputFields() output field} values.
*/
@SuppressWarnings(value = {"fallthrough"})
public static Map<FieldName, ?> evaluate(
Map<FieldName, ?> predictions, ModelEvaluationContext context) {
ModelEvaluator<?> modelEvaluator = context.getModelEvaluator();
Model model = modelEvaluator.getModel();
Output output = model.getOutput();
if (output == null) {
return predictions;
}
Map<FieldName, Object> result = new LinkedHashMap<>(predictions);
List<OutputField> outputFields = output.getOutputFields();
outputFields:
for (OutputField outputField : outputFields) {
FieldName targetFieldName = outputField.getTargetField();
Object targetValue = null;
ResultFeature resultFeature = outputField.getResultFeature();
String segmentId = outputField.getSegmentId();
SegmentResult segmentPredictions = null;
// Load the target value of the specified segment
if (segmentId != null) {
if (!(model instanceof MiningModel)) {
throw new InvalidFeatureException(outputField);
}
MiningModelEvaluationContext miningModelContext = (MiningModelEvaluationContext) context;
segmentPredictions = miningModelContext.getResult(segmentId);
// "If there is no Segment matching segmentId or if the predicate of the matching Segment
// evaluated to false, then the result delivered by this OutputField is missing"
if (segmentPredictions == null) {
continue outputFields;
} // End if
if (targetFieldName != null) {
if (!segmentPredictions.containsKey(targetFieldName)) {
throw new MissingValueException(targetFieldName, outputField);
}
targetValue = segmentPredictions.get(targetFieldName);
} else {
targetValue = segmentPredictions.getTargetValue();
}
} else
// Load the target value
{
switch (resultFeature) {
case ENTITY_ID:
{
// "Result feature entityId returns the id of the winning segment"
if (model instanceof MiningModel) {
targetValue = TypeUtil.cast(HasEntityId.class, predictions);
break;
}
}
// Falls through
default:
{
if (targetFieldName == null) {
targetFieldName = modelEvaluator.getTargetFieldName();
} // End if
if (!predictions.containsKey(targetFieldName)) {
throw new MissingValueException(targetFieldName, outputField);
}
targetValue = predictions.get(targetFieldName);
}
break;
}
}
// "If the target value is missing, then the result delivered by this OutputField is missing"
if (targetValue == null) {
continue outputFields;
}
Object value;
// Perform the requested computation on the target value
switch (resultFeature) {
case PREDICTED_VALUE:
{
value = getPredictedValue(targetValue);
}
break;
case PREDICTED_DISPLAY_VALUE:
{
DataField dataField = modelEvaluator.getDataField(targetFieldName);
if (dataField == null) {
throw new MissingFieldException(targetFieldName, outputField);
}
Target target = modelEvaluator.getTarget(targetFieldName);
value = getPredictedDisplayValue(targetValue, dataField, target);
}
break;
case TRANSFORMED_VALUE:
case DECISION:
{
if (segmentId != null) {
String name = outputField.getValue();
if (name == null) {
throw new InvalidFeatureException(outputField);
}
Expression expression = outputField.getExpression();
if (expression != null) {
throw new InvalidFeatureException(outputField);
}
value = segmentPredictions.get(FieldName.create(name));
break;
}
Expression expression = outputField.getExpression();
if (expression == null) {
throw new InvalidFeatureException(outputField);
}
value = FieldValueUtil.getValue(ExpressionUtil.evaluate(expression, context));
}
break;
case PROBABILITY:
{
value = getProbability(targetValue, outputField);
}
break;
case RESIDUAL:
{
FieldValue expectedTargetValue = context.evaluate(targetFieldName);
if (expectedTargetValue == null) {
throw new MissingValueException(targetFieldName, outputField);
}
DataField dataField = modelEvaluator.getDataField(targetFieldName);
OpType opType = dataField.getOpType();
switch (opType) {
case CONTINUOUS:
value = getContinuousResidual(targetValue, expectedTargetValue);
break;
case CATEGORICAL:
value = getCategoricalResidual(targetValue, expectedTargetValue);
break;
default:
throw new UnsupportedFeatureException(dataField, opType);
}
}
break;
case CLUSTER_ID:
{
value = getClusterId(targetValue);
}
break;
case ENTITY_ID:
{
if (targetValue instanceof HasRuleValues) {
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.RULE_ID);
break;
}
value = getEntityId(targetValue, outputField);
}
break;
case AFFINITY:
{
value = getAffinity(targetValue, outputField);
}
break;
case CLUSTER_AFFINITY:
case ENTITY_AFFINITY:
{
String entityId = outputField.getValue();
// Select the specified entity instead of the winning entity
if (entityId != null) {
value = getAffinity(targetValue, outputField);
break;
}
value = getEntityAffinity(targetValue);
}
break;
case REASON_CODE:
{
value = getReasonCode(targetValue, outputField);
}
break;
case RULE_VALUE:
{
value = getRuleValue(targetValue, outputField);
}
break;
case ANTECEDENT:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.ANTECEDENT);
}
break;
case CONSEQUENT:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.CONSEQUENT);
}
break;
case RULE:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.RULE);
}
break;
case RULE_ID:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.RULE_ID);
}
break;
case CONFIDENCE:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.CONFIDENCE);
}
break;
case SUPPORT:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.SUPPORT);
}
break;
case LIFT:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.LIFT);
}
break;
case LEVERAGE:
{
value = getRuleValue(targetValue, outputField, OutputField.RuleFeature.LEVERAGE);
}
break;
case WARNING:
{
value = context.getWarnings();
}
break;
default:
throw new UnsupportedFeatureException(outputField, resultFeature);
}
FieldValue outputValue = FieldValueUtil.create(outputField, value);
// The result of one output field becomes available to other output fields
context.declare(outputField.getName(), outputValue);
result.put(outputField.getName(), FieldValueUtil.getValue(outputValue));
}
return result;
}
private static String getClusterId(Object object) {
HasEntityId hasEntityId = TypeUtil.cast(HasEntityId.class, object);
return hasEntityId.getEntityId();
}
