/**
* Stats for data
*
* @param config
* @param statsFilePath
* @param delim
* @throws IOException
*/
public NumericalAttrStatsManager(Configuration config, String statsFilePath, String delim)
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();
Integer attr = Integer.parseInt(items[0]);
tuple.add(Tuple.STRING, items[1]);
tuple.add(Tuple.DOUBLE, items[2]);
tuple.add(Tuple.DOUBLE, items[3]);
tuple.add(Tuple.INT, items[4]);
tuple.add(Tuple.DOUBLE, items[5]);
tuple.add(Tuple.DOUBLE, items[6]);
tuple.add(Tuple.DOUBLE, items[7]);
List<Tuple> statList = stats.get(attr);
if (null == statList) {
statList = new ArrayList<Tuple>();
stats.put(attr, statList);
}
statList.add(tuple);
}
}
/**
* 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);
}
}