/**
* Build the encodings against each categorical feature.
*
* @return a list of encodings - last value of the list represent the encodings for the response
* variable. index - index of the feature value - Map<String, Integer> - map of unique values
* of this feature and there encoded values. key - unique value value - encoded value
*/
private static List<Map<String, Integer>> buildEncodings(
List<Feature> features,
Map<String, String> summaryStats,
List<Integer> newToOldIndicesList,
int responseIndex) {
List<Map<String, Integer>> encodings = new ArrayList<Map<String, Integer>>();
for (int i = 0; i < newToOldIndicesList.size() + 1; i++) {
encodings.add(new HashMap<String, Integer>());
}
for (Feature feature : features) {
Map<String, Integer> encodingMap = new HashMap<String, Integer>();
if (feature.getType().equals(FeatureType.CATEGORICAL)) {
List<String> uniqueVals =
getUniqueValues(feature.getIndex(), summaryStats.get(feature.getName()));
Collections.sort(uniqueVals);
for (int i = 0; i < uniqueVals.size(); i++) {
encodingMap.put(uniqueVals.get(i), i);
}
int newIndex = newToOldIndicesList.indexOf(feature.getIndex());
if (newIndex != -1) {
encodings.set(newIndex, encodingMap);
} else if (feature.getIndex() == responseIndex) {
// response encoding at the end
encodings.set(encodings.size() - 1, encodingMap);
}
}
}
return encodings;
}
private String getTypeOfResponseVariable(String responseVariable, List<Feature> features) {
String type = null;
for (Feature feature : features) {
if (feature.getName().equals(responseVariable)) {
type = feature.getType();
}
}
return type;
}