/**
* @param config
* @param statContent
* @param delim
* @param idOrdinals
* @param stats
* @param keyedStats
*/
private void loadMedianStatContent(
String statContent,
String delim,
int[] idOrdinals,
Map<Integer, Double> stats,
Map<String, Map<Integer, Double>> keyedStats) {
String line = null;
String[] items = null;
Scanner scanner = new Scanner(statContent);
while (scanner.hasNextLine()) {
line = scanner.nextLine();
items = line.split(delim);
if (null != idOrdinals) {
// with IDs
String compId = Utility.join(items, 0, idOrdinals.length, delim);
Map<Integer, Double> medians = keyedStats.get(compId);
if (null == medians) {
medians = new HashMap<Integer, Double>();
keyedStats.put(compId, medians);
}
medians.put(
Integer.parseInt(items[idOrdinals.length]),
Double.parseDouble(items[idOrdinals.length + 1]));
} else {
// without IDs
stats.put(Integer.parseInt(items[0]), Double.parseDouble(items[1]));
}
}
}
/**
* @param config
* @param statFilePathParam
* @param delim
* @param idOrdinals
* @param stats
* @param keyedStats
* @throws IOException
*/
private void loadMedianStat(
Configuration config,
String statFilePathParam,
String delim,
int[] idOrdinals,
Map<Integer, Double> stats,
Map<String, Map<Integer, Double>> keyedStats)
throws IOException {
List<String> lines = Utility.getFileLines(config, statFilePathParam);
for (String line : lines) {
String[] items = line.split(delim);
if (null != idOrdinals) {
// with IDs
String compId = Utility.join(items, 0, idOrdinals.length, delim);
Map<Integer, Double> medians = keyedStats.get(compId);
if (null == medians) {
medians = new HashMap<Integer, Double>();
keyedStats.put(compId, medians);
}
medians.put(
Integer.parseInt(items[idOrdinals.length]),
Double.parseDouble(items[idOrdinals.length + 1]));
} else {
// without IDs
stats.put(Integer.parseInt(items[0]), Double.parseDouble(items[1]));
}
}
}
/**
* Stats for keyed data
*
* @param config
* @param statsFilePath
* @param delim
* @param idOrdinals
* @throws IOException
*/
public NumericalAttrStatsManager(
Configuration config, String statsFilePath, String delim, int[] idOrdinals)
throws IOException {
InputStream fs = Utility.getFileStream(config, statsFilePath);
BufferedReader reader = new BufferedReader(new InputStreamReader(fs));
String line = null;
String[] items = null;
// (0)attr ord (1)cond attr (2)sum (3)sum square (4)count (5)mean (6)variance (7)std dev (8)min
// (9)max
while ((line = reader.readLine()) != null) {
items = line.split(delim);
Tuple tuple = new Tuple();
int i = 0;
String compKey = Utility.join(items, 0, idOrdinals.length);
i += idOrdinals.length;
Integer attr = Integer.parseInt(items[i++]);
tuple.add(Tuple.STRING, items[i++]);
tuple.add(Tuple.DOUBLE, items[i++]);
tuple.add(Tuple.DOUBLE, items[i++]);
tuple.add(Tuple.INT, items[i++]);
tuple.add(Tuple.DOUBLE, items[i++]);
tuple.add(Tuple.DOUBLE, items[i++]);
tuple.add(Tuple.DOUBLE, items[i++]);
// add to map
Map<Integer, List<Tuple>> stats = keyedStats.get(compKey);
if (null == stats) {
stats = new HashMap<Integer, List<Tuple>>();
keyedStats.put(compKey, stats);
}
List<Tuple> statList = stats.get(attr);
if (null == statList) {
statList = new ArrayList<Tuple>();
stats.put(attr, statList);
}
statList.add(tuple);
}
}