001 /*
002 * Created on Apr 9, 2005
003 *
004 */
005 package aima.learning.framework;
006
007 /**
008 * @author Ravi Mohan
009 *
010 */
011 import java.io.BufferedReader;
012 import java.io.InputStreamReader;
013 import java.util.Arrays;
014 import java.util.Hashtable;
015 import java.util.Iterator;
016 import java.util.List;
017
018 import aima.util.Util;
019
020 public class DataSetFactory {
021
022 public DataSet fromFile(String filename, DataSetSpecification spec,
023 String separator) throws Exception {
024 // assumed file in data directory and ends in .csv
025 DataSet ds = new DataSet(spec);
026 BufferedReader reader = new BufferedReader(
027 new InputStreamReader(this.getClass().getResourceAsStream(
028 "../data/" + filename + ".csv")));
029 String line;
030 while ((line = reader.readLine()) != null) {
031 ds.add(exampleFromString(line, spec, separator));
032 }
033
034 return ds;
035
036 }
037
038 public static Example exampleFromString(String data,
039 DataSetSpecification dataSetSpec, String separator) {
040 Hashtable<String, Attribute> attributes = new Hashtable<String, Attribute>();
041 List<String> attributeValues = Arrays.asList(data.split(separator));
042 if (dataSetSpec.isValid(attributeValues)) {
043 List<String> names = dataSetSpec.getAttributeNames();
044 Iterator<String> nameiter = names.iterator();
045 Iterator<String> valueiter = attributeValues.iterator();
046 while (nameiter.hasNext() && valueiter.hasNext()) {
047 String name = nameiter.next();
048 AttributeSpecification attributeSpec = dataSetSpec
049 .getAttributeSpecFor(name);
050 Attribute attribute = attributeSpec.createAttribute(valueiter
051 .next());
052 attributes.put(name, attribute);
053 }
054 String targetAttributeName = dataSetSpec.getTarget();
055 AttributeSpecification attributeSpec = dataSetSpec
056 .getAttributeSpecFor(targetAttributeName);
057 return new Example(attributes, attributes.get(targetAttributeName));
058 } else {
059 throw new RuntimeException("Unable to construct Example from "
060 + data);
061 }
062 }
063
064 public static DataSet getRestaurantDataSet() throws Exception {
065 DataSetSpecification spec = createRestaurantDataSetSpec();
066 return new DataSetFactory().fromFile("restaurant", spec, "\\s+");
067 }
068
069 public static DataSetSpecification createRestaurantDataSetSpec() {
070 DataSetSpecification dss = new DataSetSpecification();
071 dss.defineStringAttribute("alternate", Util.yesno());
072 dss.defineStringAttribute("bar", Util.yesno());
073 dss.defineStringAttribute("fri/sat", Util.yesno());
074 dss.defineStringAttribute("hungry", Util.yesno());
075 dss.defineStringAttribute("patrons", new String[] { "None", "Some",
076 "Full" });
077 dss.defineStringAttribute("price", new String[] { "$", "$$", "$$$" });
078 dss.defineStringAttribute("raining", Util.yesno());
079 dss.defineStringAttribute("reservation", Util.yesno());
080 dss.defineStringAttribute("type", new String[] { "French", "Italian",
081 "Thai", "Burger" });
082 dss.defineStringAttribute("wait_estimate", new String[] { "0-10",
083 "10-30", "30-60", ">60" });
084 dss.defineStringAttribute("will_wait", Util.yesno());
085 // last attribute is the target attribute unless the target is
086 // explicitly reset with dss.setTarget(name)
087
088 return dss;
089 }
090
091 public static DataSet getIrisDataSet() throws Exception {
092 DataSetSpecification spec = createIrisDataSetSpec();
093 return new DataSetFactory().fromFile("iris", spec, ",");
094 }
095
096 public static DataSetSpecification createIrisDataSetSpec() {
097 DataSetSpecification dss = new DataSetSpecification();
098 dss.defineNumericAttribute("sepal_length");
099 dss.defineNumericAttribute("sepal_width");
100 dss.defineNumericAttribute("petal_length");
101 dss.defineNumericAttribute("petal_width");
102 dss.defineStringAttribute("plant_category", new String[] { "setosa",
103 "versicolor", "virginica" });
104 return dss;
105 }
106
107 }