Class VariableEncodings

src/com/winvector/variables/VariableEncodings.java:15–163

import com.winvector.variables.LevelVectors.VectorRow;

public final class VariableEncodings implements Serializable {
	private static final long serialVersionUID = 1L;
	
	// definitional 
	private final PrimaVariableInfo def;
	private final boolean useIntercept;
	// weight control
	public final String weightKey;
	// derived inputs
	public final int vdim;
	public final ArrayList<VariableMapping> adaptions = new ArrayList<VariableMapping>();
	// derived results
	public final SortedMap<String,Integer> outcomeCategories = new TreeMap<String,Integer>();
	private final ArrayList<String> outcomeNames = new ArrayList<String>();
	
	public VariableEncodings(final PrimaVariableInfo def, final boolean useIntercept, final String weightKey,
			final Map<String,Map<String,VectorRow>> vectorEncodings) {
		this.def = def;
		this.useIntercept = useIntercept;
		this.weightKey = weightKey;
		// encode variables
		int adapterDim = 0;
		if(useIntercept) {
			final VariableMapping adaption = new ConstRecord(adapterDim);
			adaptions.add(adaption);
			adapterDim += adaption.indexR() - adaption.indexL();
		}
		for(final String ni: new TreeSet<String>(def.numericColumnSet.keySet())) {
			final VariableMapping adaption = new NumericVariable(ni,adapterDim);
			adaptions.add(adaption);
			adapterDim += adaption.indexR() - adaption.indexL();
		}
		//final Random rand = new Random(15135);
		for(final String ci: new TreeSet<String>(def.catLevels.keySet())) {
			final VariableMapping adaption;
			if((vectorEncodings!=null)&&(vectorEncodings.containsKey(ci))) {
				adaption = new LevelVectors(ci,adapterDim,vectorEncodings.get(ci));
			} else {
				final CountMap<String> levels = def.catLevels.get(ci);
				adaption = new LevelIndicators(ci,adapterDim,levels.keySet());
			}
			adaptions.add(adaption);
			adapterDim += adaption.indexR() - adaption.indexL();
		}
		vdim = adapterDim;
		// encode outcome
		for(final String oci: new TreeSet<String>(def.outcomes.keySet())) {
			outcomeCategories.put(oci,outcomeCategories.size());
			outcomeNames.add(oci);
		}
	}

	public VariableEncodings(final PrimaVariableInfo def, final boolean useIntercept, final String weightKey) {
		this(def,useIntercept,weightKey,null);
	}
	
	public PrimaVariableInfo def() {
		return def;