void RfExtractorClassifier::import(istream& is)
{
_getExtractors();
_rf.reset(new RandomForest());
_rf->import(is);
vector<string> factorLabels = _rf->getFactorLabels();
QStringList extractorNames;
for (size_t i = 0; i < _extractors.size(); i++)
{
extractorNames.append(QString::fromUtf8(_extractors[i]->getName().data()).
replace(QRegExp("[^\\w]"), "_"));
}
QStringList missingExtractors;
for (size_t i = 0; i < factorLabels.size(); i++)
{
QString fn = QString::fromUtf8(factorLabels[i].data());
if (extractorNames.contains(fn) == false)
{
missingExtractors.append(fn);
}
_rfFactorLabels.append(fn);
}
if (missingExtractors.size() > 0)
{
LOG_WARN("An extractor used by the model is not being calculated. We will still try, but");
LOG_WARN("this will undoubtably result in poor quality matches. Missing extractors:");
LOG_WARN(missingExtractors);
LOG_WARN("Available extractors: " << extractorNames);
}
}
void RfExtractorClassifier::import(QDomElement& docRoot)
{
_getExtractors();
_rf.reset(new RandomForest());
_rf->importModel(docRoot);
vector<string> factorLabels = _rf->getFactorLabels();
QStringList extractorNames;
for (size_t i = 0; i < _extractors.size(); i++)
{
extractorNames.append(QString::fromUtf8(_extractors[i]->getName().data()).
replace(QRegExp("[^\\w]"), "_"));
}
QStringList missingExtractors;
for (size_t i = 0; i < factorLabels.size(); i++)
{
QString fn = QString::fromUtf8(factorLabels[i].data());
if (extractorNames.contains(fn) == false)
{
missingExtractors.append(fn);
}
_rfFactorLabels.append(fn);
}
if (missingExtractors.size() > 0)
{
if (logWarnCount < Log::getWarnMessageLimit())
{
LOG_WARN(
"An extractor used by the model is not being calculated. We will still try, but this will "
"undoubtably result in poor quality matches. Missing extractors:");
}
else if (logWarnCount == Log::getWarnMessageLimit())
{
LOG_WARN(className() << ": " << Log::LOG_WARN_LIMIT_REACHED_MESSAGE);
}
logWarnCount++;
LOG_TRACE("Missing extractors: " << missingExtractors);
LOG_TRACE("Available extractors: " << extractorNames);
}
}
map<QString, double> RfExtractorClassifier::getFeatures(const shared_ptr<const OsmMap>& m,
ElementId eid1, ElementId eid2) const
{
map<QString, double> result;
const shared_ptr<const Element>& e1 = m->getElement(eid1);
const shared_ptr<const Element> e2 = m->getElement(eid2);
_getExtractors();
for (size_t i = 0; i < _extractors.size(); i++)
{
double v = _extractors[i]->extract(*m, e1, e2);
// if it isn't null then include it.
if (!FeatureExtractor::isNull(v))
{
QString factorName = QString::fromStdString(_extractors[i]->getName()).
replace(QRegExp("[^\\w]"), "_");
result[factorName] = v;
}
}
return result;
}
