@SuppressWarnings("unchecked")
public double crossValidationError(int n, Approach approach) throws IOException {
List<String> queries = QueryLoader.loadQueries(queryFile);
int partSize = queries.size() / n;
List<List<String>> partitions = new ArrayList<>(n);
for (int i = 0; i < (n - 1); i++) {
partitions.add(queries.subList(i * partSize, (i + 1) * partSize));
}
partitions.add(queries.subList((n - 1) * partSize, queries.size()));
Model model = readModel(MODEL_FILE);
LOGGER.info("Generating expected counts...");
ObjectIntOpenHashMap<String> gsResults[] = countResources(partitions, model);
double rootMeanSquareSum = 0;
double foldErrors[] = new double[n];
List<String> training, predicted;
for (int i = 0; i < n; i++) {
LOGGER.info("Starting fold " + i + "...");
training = generateTrainingSet(i, partitions);
predicted = approach.generateResourceRanking(training, model);
foldErrors[i] = RMSD.getRMSD(predicted, generateExpectedResult(i, gsResults));
LOGGER.info("Error of fold " + i + " = " + foldErrors[i]);
rootMeanSquareSum += foldErrors[i];
}
LOGGER.info("Error of folds " + Arrays.toString(foldErrors));
return rootMeanSquareSum / n;
}
// constructor
public TQuery(String[] args) {
try {
long startTime = System.currentTimeMillis();
processArgs(args);
queryIndex = new QueryIndex(this);
queryLoader = new QueryLoader(this);
// print some stats on building the engine
String diffTime =
Num.formatNumberOneFraction(((double) (System.currentTimeMillis() - startTime)) / 1000);
int numFiles = vcfDataFiles.length + bedDataFiles.length + mafDataFiles.length;
System.err.println("\n" + diffTime + " Sec to build using " + IO.memory() + " of RAM");
System.err.println("\t" + numFiles + "\tData sources loaded");
System.err.println("\t" + dataSources.getRecordsLoaded() + "\tRecords indexed");
System.err.println("\t" + dataSources.getRecordsSkipped() + "\tRecords skipped\n");
// print summary of available filters
System.err.println(dataSources.fetchSummary());
queryFilesFromCmdLine();
// release file handles
queryLoader.closeTabixReaders();
} catch (Exception e) {
e.printStackTrace();
System.err.println("\nProblem with executing the TQuery!");
}
}
public double validate(Approach approach, Model model, QueryExecutor executor)
throws IOException {
List<String> queries = QueryLoader.loadQueries(queryFile);
Baseline baseline = new Baseline(executor);
return RMSD.getRMSD(
approach.generateResourceRanking(queries, model),
generateUriRankRangeMapping(baseline.sumUpResults(model, queries)));
}
