@Override
protected List<TTDataSet> perform() throws TaskExecutionException, InterruptedException {
Preconditions.checkNotNull(sources);
List<TTDataSet> datasets = new ArrayList<TTDataSet>(sources.size());
for (TTDataSet dataset : sources) {
try {
GenericTTDataBuilder builder = new GenericTTDataBuilder();
if (getName() == null) {
builder.setName(dataset.getName());
} else {
builder.setName(getName());
}
for (Map.Entry<String, Object> entry : dataset.getAttributes().entrySet()) {
builder.setAttribute(entry.getKey(), entry.getValue());
}
builder.setAttribute("Retain", retain);
builder.setQuery(dataset.getQueryData());
builder.setTest(dataset.getTestData());
builder.setTrain(
downsample(dataset.getTrainingData(), dataset.getTestData().getUserDAO().getUserIds()));
datasets.add(builder.build());
} catch (IOException e) {
throw new TaskExecutionException(e);
}
}
return datasets;
}
/**
* Prefix a table for a particular algorithmInfo and data set.
*
* @param base The table to prefix.
* @param algorithm The algorithmInfo to prefix for.
* @param dataSet The data set to prefix for.
* @return A prefixed table, suitable for outputting the results of evaluating {@code
* algorithmInfo} on {@code dataSet}, or {@code null} if {@code base} is null.
*/
public TableWriter prefixTable(TableWriter base, Attributed algorithm, TTDataSet dataSet) {
if (base == null) {
return null;
}
Object[] prefix = new Object[getCommonColumnCount()];
prefix[0] = algorithm.getName();
for (Map.Entry<String, Object> attr : dataSet.getAttributes().entrySet()) {
int idx = getDataColumn(attr.getKey());
prefix[idx] = attr.getValue();
}
for (Map.Entry<String, Object> attr : algorithm.getAttributes().entrySet()) {
int idx = getAlgorithmColumn(attr.getKey());
prefix[idx] = attr.getValue();
}
return TableWriters.prefixed(base, prefix);
}