Class PrimaVariableInfo

src/com/winvector/variables/PrimaVariableInfo.java:16–187

import com.winvector.util.StatMap;

public class PrimaVariableInfo implements Serializable {
	private static final long serialVersionUID = 1L;
	
	// definitional
	public String resultColumn;
	public SortedSet<String> variables = new TreeSet<String>();
	public SortedSet<String> forcedNumeric = new TreeSet<String>();
	public SortedSet<String> forcedCategorical = new TreeSet<String>();
	// calculated
	public CountMap<String> outcomes = new CountMap<String>(CountMap.strCmp);
	public StatMap numericColumnSet = new StatMap();
	public SortedMap<String,CountMap<String>> catLevels = new TreeMap<String,CountMap<String>>();

	protected void readyForDefTracking() {
		outcomes = new CountMap<String>(CountMap.strCmp);
		numericColumnSet = new StatMap();
		catLevels = new TreeMap<String,CountMap<String>>();
		for(final String nc: forcedNumeric) {
			numericColumnSet.observe(nc,0.0,0.0);
		}
		for(final String key: forcedCategorical) {
			catLevels.put(key,new CountMap<String>(CountMap.strCmp));
		}		
	}
	
	public void readyForDefTracking(final Formula formula) {
		resultColumn = formula.resultColumn;
		variables = new TreeSet<String>(formula.variables);
		forcedNumeric = new TreeSet<String>(formula.forcedNumeric);
		forcedCategorical = new TreeSet<String>(formula.forcedCategorical);
		readyForDefTracking();
	}

	/**
	 * confirm row has complete set of non-empty independent variables
	 * @param row
	 * @return
	 */
	public boolean completeSetOfVars(final BurstMap row) {
		for(final String key: variables) {
			final String sValue = row.getAsString(key);
			if((sValue==null)||(sValue.trim().length()<=0)) {
				return false;
			}
		}
		return true;
	}
		
	/**
	 * track: sets of levels, mean of numeric variable and distribution of outcomes
	 * @param row
	 */
	public void trackVariableDefsFromRow(final BurstMap row) {
		if(!completeSetOfVars(row)) {
			return;
		}
		for(final String key: variables) {
			if(catLevels.containsKey(key)||forcedNumeric.contains(key)) {