/**
* Samples n ids based on their a Table which contains weights, probabilities or frequencies.
*
* @param weightedTable
* @param n
* @param withReplacement
* @return
*/
public static FlatDataCollection weightedSampling(
AssociativeArray weightedTable, int n, boolean withReplacement) {
FlatDataList sampledIds = new FlatDataList();
double sumOfFrequencies = Descriptives.sum(weightedTable.toFlatDataCollection());
int populationN = weightedTable.size();
for (int i = 0; i < n; ++i) {
if (withReplacement == false && populationN <= n) {
// if replacement is not allowed and we already sampled everything that it can stop
break;
}
double randomFrequency = PHPMethods.mt_rand(0.0, sumOfFrequencies);
double cumulativeFrequency = 0;
for (Map.Entry<Object, Object> entry : weightedTable.entrySet()) {
Object pointID = entry.getKey();
cumulativeFrequency += TypeInference.toDouble(entry.getValue());
if (cumulativeFrequency >= randomFrequency) {
if (withReplacement == false) {
/* if replacement is not allowed check if the point already exists */
if (sampledIds.contains(pointID)) {
continue;
}
}
sampledIds.add(pointID);
break;
}
}
}
return sampledIds.toFlatDataCollection();
}
/**
* Calculate the variance from the sample
*
* @param flatDataCollection
* @return
*/
public static double variance(FlatDataCollection flatDataCollection) {
return Descriptives.variance(flatDataCollection, true);
}
/**
* Calculate the mean from the sample
*
* @param flatDataCollection
* @return
*/
public static double mean(FlatDataCollection flatDataCollection) {
return Descriptives.mean(flatDataCollection);
}