wfolsom/cs6601p2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Work in progress.

Browse Source 
Able to read short survey, recipe book but crashing in Bayes Net code.
This commit is contained in:
Woody Folsom
2012-03-11 21:25:04 -04:00
parent 3046f68681
commit bb94356ec1
11 changed files with 999 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

218
src/dkohl/bayes/builders/FoodNetBuilder.java Normal file
View File

@@ -0,0 +1,218 @@
package dkohl.bayes.builders;
import java.util.HashSet;
import net.woodyfolsom.cs6601.p2.Diner;
import net.woodyfolsom.cs6601.p2.Ingredient.TYPE;
import net.woodyfolsom.cs6601.p2.Ingredients;
import net.woodyfolsom.cs6601.p2.Recipe;
import net.woodyfolsom.cs6601.p2.RecipeBook;
import net.woodyfolsom.cs6601.p2.Survey;
import dkohl.bayes.bayesnet.BayesNet;
import dkohl.bayes.estimation.MaximumLikelihoodEstimation;
import dkohl.bayes.example.builders.FoodExampleBuilder;
import dkohl.bayes.probability.Assignment;
import dkohl.bayes.probability.Probability;
import dkohl.bayes.probability.Variable;
import dkohl.bayes.probability.distribution.ContinousDistribution;
import dkohl.bayes.probability.distribution.ProbabilityDistribution;
import dkohl.bayes.probability.distribution.ProbabilityTable;
import dkohl.bayes.statistic.DataPoint;
import dkohl.bayes.statistic.DataSet;
import dkohl.onthology.Ontology;
public class FoodNetBuilder {
public static final String TASTE = "Taste";
public static final String SOMEONE_VEGETARIAN = "Vegetarian";
public static final String CONTAINS_MEAT = "Meat";
public static final String CONTAINS_VEGETABLE = "Vegetable";
public static final String CONTAINS_BEEF = TYPE.BEEF.toString();
public static final String CONTAINS_PORK = TYPE.PORK.toString();
public static final String CONTAINS_TOMATOS = TYPE.TOMATO.toString();
public static final String CONTAINS_POTATOS = TYPE.POTATO.toString();
public static final String TRUE_VALUE = "true";
public static final String FALSE_VALUE = "false";
public static final String DOMAIN[] = { TRUE_VALUE, FALSE_VALUE };
public static final String RATING_DOMAIN[] = { "1", "2", "3", "4", "5",
"6", "7", "8", "9", "10" };
private static final String[] VARIABLES = { SOMEONE_VEGETARIAN,
CONTAINS_BEEF, CONTAINS_MEAT, CONTAINS_PORK, CONTAINS_POTATOS,
CONTAINS_TOMATOS, CONTAINS_VEGETABLE, TASTE };
private static final String[] OBSERVED = { CONTAINS_BEEF, CONTAINS_PORK,
CONTAINS_POTATOS, CONTAINS_TOMATOS, };
public static Ontology createOntology() {
HashSet<String> classes = new HashSet<String>();
classes.add(CONTAINS_MEAT);
classes.add(CONTAINS_VEGETABLE);
Ontology onthology = new Ontology(classes);
onthology.define(CONTAINS_PORK, CONTAINS_MEAT);
onthology.define(CONTAINS_BEEF, CONTAINS_MEAT);
onthology.define(CONTAINS_TOMATOS, CONTAINS_VEGETABLE);
onthology.define(CONTAINS_POTATOS, CONTAINS_VEGETABLE);
return onthology;
}
private static Assignment build(String varible, String value) {
return new Assignment(new Variable(varible, DOMAIN), value);
}
public static DataPoint normalize(DataPoint point, Ontology onto) {
// resolve onthology
DataPoint normPoint = new DataPoint(point);
for (String key : point.keySet()) {
if (onto.getInheritance().containsKey(key)) {
normPoint
.add(build(onto.getInheritance().get(key), TRUE_VALUE));
}
}
// implement closed world assumption
// everything unknown is false
for (String variable : VARIABLES) {
if (!normPoint.containsKey(variable)) {
normPoint.add(build(variable, FALSE_VALUE));
}
}
return normPoint;
}
public static DataSet getSurveyDataSet(Survey survey, RecipeBook recipeBook) {
DataSet data = FoodExampleBuilder.examples();
Ontology onto = createOntology();
int nDishes = survey.getDishCount();
for (int dinerIndex = 0; dinerIndex < survey.getDinerCount(); dinerIndex++) {
Diner diner = survey.getDiner(dinerIndex);
for (int dishIndex = 0; dishIndex < nDishes; dishIndex++) {
data.add(normalize(createDataPoint(recipeBook, survey.getDish(dishIndex), diner.getRating(dishIndex)),onto));
}
}
return data;
}
public static DataPoint createDataPoint(RecipeBook recipeBook, String recipeName, int weight) {
DataPoint point = new DataPoint();
Recipe recipe = recipeBook.getRecipe(recipeName);
Ingredients ingredients = recipe.getIngredients();
if (ingredients.contains(TYPE.BEEF)) {
point.add(build(CONTAINS_BEEF, TRUE_VALUE));
}
if (ingredients.contains(TYPE.PORK)) {
point.add(build(CONTAINS_PORK, TRUE_VALUE));
}
if (ingredients.contains(TYPE.POTATO)) {
point.add(build(CONTAINS_POTATOS, TRUE_VALUE));
}
if (ingredients.contains(TYPE.TOMATO)) {
point.add(build(CONTAINS_TOMATOS, TRUE_VALUE));
}
point.add(build(TASTE, "" + weight));
return point;
}
public static ProbabilityDistribution vegi() {
String names[] = { SOMEONE_VEGETARIAN };
ProbabilityTable table = new ProbabilityTable(names);
table.setProbabilityForAssignment("true;", new Probability(0));
table.setProbabilityForAssignment("false;", new Probability(1));
return table;
}
public static ProbabilityDistribution beef() {
String names[] = { CONTAINS_MEAT, CONTAINS_BEEF };
ProbabilityTable table = new ProbabilityTable(names);
return table;
}
public static ProbabilityDistribution pork() {
String names[] = { CONTAINS_MEAT, CONTAINS_PORK };
ProbabilityTable table = new ProbabilityTable(names);
return table;
}
public static ProbabilityDistribution meet() {
String names[] = { SOMEONE_VEGETARIAN, CONTAINS_MEAT };
ProbabilityTable table = new ProbabilityTable(names);
return table;
}
public static ProbabilityDistribution tomatos() {
String names[] = { CONTAINS_VEGETABLE, CONTAINS_TOMATOS };
ProbabilityTable table = new ProbabilityTable(names);
return table;
}
public static ProbabilityDistribution potatos() {
String names[] = { CONTAINS_VEGETABLE, CONTAINS_POTATOS };
ProbabilityTable table = new ProbabilityTable(names);
return table;
}
public static ProbabilityDistribution vegetables() {
String names[] = { CONTAINS_VEGETABLE, SOMEONE_VEGETARIAN };
ProbabilityTable table = new ProbabilityTable(names);
return table;
}
public static ProbabilityDistribution taste() {
String names[] = { TASTE, CONTAINS_BEEF, CONTAINS_PORK,
CONTAINS_POTATOS, CONTAINS_TOMATOS };
ContinousDistribution distribution = new ContinousDistribution(names, 0);
return distribution;
}
public static BayesNet createDishNet(Survey survey, RecipeBook recipeBook) {
BayesNet net = new BayesNet(VARIABLES);
net.setDistribution(new Variable(SOMEONE_VEGETARIAN, DOMAIN), vegi());
net.setDistribution(new Variable(CONTAINS_MEAT, DOMAIN), meet());
net.setDistribution(new Variable(CONTAINS_VEGETABLE, DOMAIN),
vegetables());
net.setDistribution(new Variable(CONTAINS_BEEF, DOMAIN), beef());
net.setDistribution(new Variable(CONTAINS_PORK, DOMAIN), pork());
net.setDistribution(new Variable(CONTAINS_POTATOS, DOMAIN), potatos());
net.setDistribution(new Variable(CONTAINS_TOMATOS, DOMAIN), tomatos());
net.setDistribution(new Variable(TASTE, RATING_DOMAIN), taste());
Ontology ontology = createOntology();
for (String category : ontology.getClasses()) {
net.connect(category, SOMEONE_VEGETARIAN);
}
for (String thing : OBSERVED) {
net.connect(thing, ontology.getInheritance().get(thing));
net.connect(TASTE, thing);
}
DataSet dataSet = getSurveyDataSet(survey, recipeBook);
for (String category : ontology.getClasses()) {
MaximumLikelihoodEstimation.estimate(dataSet, net, category);
for (String thing : ontology.getClasses2thing().get(category)) {
MaximumLikelihoodEstimation.estimate(dataSet, net, thing);
}
}
MaximumLikelihoodEstimation.estimate(dataSet, net, TASTE);
ContinousDistribution distribturion = (ContinousDistribution) net
.getNodes().get(TASTE);
distribturion.estimate();
return net;
}
}

4
src/dkohl/bayes/inference/EnumerateAll.java
View File

@@ -27,8 +27,8 @@ public class EnumerateAll {
* a set of assignments in this net.
* @return
*/
public static LinkedList<ProbabilityAssignment> enumerateAsk(
Variable query, BayesNet net, LinkedList<Assignment> assignments) {
public static List<ProbabilityAssignment> enumerateAsk(
Variable query, BayesNet net, List<Assignment> assignments) {
LinkedList<Variable> variables = net.getVariables();
LinkedList<ProbabilityAssignment> result = new LinkedList<ProbabilityAssignment>();

89
src/net/woodyfolsom/cs6601/p2/BayesChef.java
View File

@@ -1,61 +1,112 @@
package net.woodyfolsom.cs6601.p2;
import java.io.File;
import java.util.LinkedList;
import java.util.List;
import org.apache.commons.math3.stat.regression.SimpleRegression;
import net.woodyfolsom.cs6601.p2.Ingredient.TYPE;
import dkohl.bayes.bayesnet.BayesNet;
import dkohl.bayes.builders.FoodNetBuilder;
import dkohl.bayes.example.builders.FoodExampleBuilder;
import dkohl.bayes.inference.EnumerateAll;
import dkohl.bayes.probability.Assignment;
import dkohl.bayes.probability.ProbabilityAssignment;
import dkohl.bayes.probability.Variable;
public class BayesChef {
public static void main(String... args) {
System.out.println("Reading recipe book.");
RecipeBook recipeBook = RecipeBookReader.readRecipeBook(new File("data/survey_recipes.xml"));
System.out.println("Read data for " + recipeBook.getSize() + " recipes.");
RecipeBook recipeBook = RecipeBookReader.readRecipeBook(new File("data/short_recipebook.xml"));
System.out.println("Reading survey data.");
Survey survey = SurveyReader.readSurvey(new File("data/survey.xml"));
Survey survey = SurveyReader.readSurvey(new File("data/short_survey.xml"));
System.out.println("Read data for " + survey.getDinerCount() + " diner(s).");
new BayesChef().reportBestMeal(recipeBook,survey);
}
private void reportBestMeal(RecipeBook recipeBook, Survey survey) {
int numRecipes = recipeBook.getSize();
System.out.println("Read data for " + numRecipes + " recipes.");
int numDiners = survey.getDinerCount();
System.out.println("Read data for " + numDiners + " diner(s).");
/*
System.out.println("Setting evidence for first 11 recipes.");
System.out.println("Evaluating preference for remaining recipes.");
//read survey prefs from survey.xml
double[][] surveyPrefs = new double[5][];
for (int i = 0; i < 5; i++) {
surveyPrefs[i] = new double[11];
for (int j = 0; j < 11; j++) {
double[][] surveyPrefs = new double[numDiners][];
for (int i = 0; i < numDiners; i++) {
surveyPrefs[i] = new double[numRecipes/2];
for (int j = 0; j < numRecipes/2; j++) {
surveyPrefs[i][j] = survey.getDiner(i).getRating(j);
}
}
//generating stub evaluated preferences to test RMSE calculation
double[][] calculatedPrefs = new double[5][];
calculatedPrefs[0] = new double[] { 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 };
calculatedPrefs[1] = new double[] { 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 };
calculatedPrefs[2] = new double[] { 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 };
calculatedPrefs[3] = new double[] { 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 };
calculatedPrefs[4] = new double[] { 10.0, 10.0, 0.0, 10.0, 10.0, 4.0, 10.0, 10.0, 10.0, 10.0, 10.0 };
double[][] calculatedPrefs = new double[numDiners][];
calculatedPrefs[0] = new double[numRecipes/2];
System.out.println("RMSE for recipes #12-22 (calculated vs. surveyed preference):");
System.out.println("RMSE for recipes #" + numRecipes/2 + "-" + numRecipes +" (calculated vs. surveyed preference):");
for (int dinerIndex = 0; dinerIndex < survey.getDinerCount(); dinerIndex++) {
SimpleRegression simpleRegression = new SimpleRegression();
for (int i = 0; i < 11; i++) {
for (int i = 0; i < numRecipes/2; i++) {
simpleRegression.addData(i,calculatedPrefs[dinerIndex][i]);
}
double calculatedMSE = simpleRegression.getMeanSquareError();
simpleRegression.clear();
for (int i = 0; i < 11; i++) {
for (int i = 0; i < numRecipes/2; i++) {
simpleRegression.addData(i,surveyPrefs[dinerIndex][i]);
}
double surveyMSE = simpleRegression.getMeanSquareError();
System.out.println("Diner # " + (dinerIndex + 1) + ": " + calculatedMSE + " vs. "+ surveyMSE);
}*/
BayesNet net = FoodNetBuilder.createDishNet(survey, recipeBook);
//BayesNet net = FoodExampleBuilder.dishNet();
printPreference(net, TYPE.PORK);
printPreference(net, TYPE.BEEF);
printPreference(net, TYPE.POTATO);
printPreference(net, TYPE.TOMATO);
}
void printPreference(BayesNet net, TYPE type) {
List<ProbabilityAssignment> probs = EnumerateAll.enumerateAsk(new Variable(
FoodNetBuilder.TASTE, FoodNetBuilder.RATING_DOMAIN),
net, createQuery(type));
double max_val = 0.0;
String max_arg = "N/A";
for (ProbabilityAssignment p : probs) {
if (p.getProbability() > max_val) {
max_val = p.getProbability();
max_arg = p.getValue();
}
}
System.out.println("TASTE " + type + ": " + max_arg + " " + max_val);
}
List<Assignment> createQuery(TYPE type) {
List<Assignment> assignment = new LinkedList<Assignment>();
switch (type) {
case PORK:
case BEEF:
case POTATO:
case TOMATO:
assignment.add(new Assignment(new Variable(type.toString(), FoodNetBuilder.DOMAIN),
FoodNetBuilder.TRUE_VALUE));
break;
default:
}
return assignment;
}
}

17
src/net/woodyfolsom/cs6601/p2/Ingredient.java
View File

@@ -4,7 +4,7 @@ import com.thoughtworks.xstream.annotations.XStreamAlias;
@XStreamAlias("ing")
public class Ingredient {
public enum TYPE { ALCOHOL, DAIRY, EGGS, FISH, GLUTEN, GRAIN, NUTS, PORK, POULTRY, RED_MEAT, SHELLFISH, SPICE, SUGAR}
public enum TYPE { ALCOHOL, BEEF, DAIRY, EGGS, FISH, GLUTEN, GRAIN, NUTS, PORK, POULTRY, POTATO, SHELLFISH, SPICE, SUGAR, TOMATO}
private String item;
@@ -14,20 +14,31 @@ public class Ingredient {
public boolean isType(TYPE type) {
switch (type) {
case BEEF :
return item.contains("beef");
case DAIRY :
return item.contains("margarine");
case EGGS :
return item.equals("egg") || item.equals("eggs");
case GLUTEN :
return item.contains("flour");
case RED_MEAT :
return item.contains("beef");
case PORK :
return item.contains("pork");
case POTATO :
return item.contains("potato");
case SPICE :
return item.endsWith("cinnamon") || item.endsWith("nutmeg") || item.endsWith("cloves");
case SUGAR :
return item.endsWith("sugar");
case TOMATO :
return item.contains("tomato");
default : //unknown ingredient, e.g. coffee, bananas, honey
return false;
}
}
public Object readResolve() {
item = item.toLowerCase();
return this;
}
}

10
src/net/woodyfolsom/cs6601/p2/RecipeBook.java
View File

@@ -15,6 +15,16 @@ public class RecipeBook {
return recipes.get(index);
}
//TODO build an index of recipes by name?
public Recipe getRecipe(String recipeName) {
for (Recipe recipe : recipes) {
if (recipeName.equalsIgnoreCase(recipe.getHead().getTitle())) {
return recipe;
}
}
return null;
}
public int getSize() {
return recipes.size();
}