public void testDLTestMatchSucceedsWithMatchedExample()
{
DataSet ds = DataSetFactory.getRestaurantDataSet();
Example e = ds.getExample(0);
tvn.cosine.ai.learning.inductive.DecisionListTest test = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test.add("type", "French");
Assert.IsTrue(test.matches(e));
}
public void testDLTestMatchFailsOnMismatchedExample()
{
DataSet ds = DataSetFactory.getRestaurantDataSet();
Example e = ds.getExample(0);
tvn.cosine.ai.learning.inductive.DecisionListTest test = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test.add("type", "Thai");
Assert.IsFalse(test.matches(e));
}
public void testDecisionListHandlesEmptyDataSet()
{
// tests first base case of recursion
DecisionList dlist = new DecisionList("Yes", "No");
tvn.cosine.ai.learning.inductive.DecisionListTest test1 = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test1.add("type", "Thai"); // doesn't match first example
dlist.add(test1, "test1success");
}
public void testDecisionListWithSingleTestReturnsTestValueIfTestSuccessful()
{
DecisionList dlist = new DecisionList("Yes", "No");
DataSet ds = DataSetFactory.getRestaurantDataSet();
tvn.cosine.ai.learning.inductive.DecisionListTest test = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test.add("type", "French");
dlist.add(test, "test1success");
Assert.AreEqual("test1success", dlist.predict(ds.getExample(0)));
}
public void testDLTestReturnsMatchedAndUnmatchedExamplesCorrectly()
{
DataSet ds = DataSetFactory.getRestaurantDataSet();
tvn.cosine.ai.learning.inductive.DecisionListTest test = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test.add("type", "Burger");
DataSet matched = test.matchedExamples(ds);
Assert.AreEqual(4, matched.size());
DataSet unmatched = test.unmatchedExamples(ds);
Assert.AreEqual(8, unmatched.size());
}
public void testDecisionListFallsThruToDefaultIfNoTestMatches()
{
DecisionList dlist = new DecisionList("Yes", "No");
DataSet ds = DataSetFactory.getRestaurantDataSet();
tvn.cosine.ai.learning.inductive.DecisionListTest test1 = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test1.add("type", "Thai"); // doesn't match first example
dlist.add(test1, "test1success");
tvn.cosine.ai.learning.inductive.DecisionListTest test2 = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test2.add("type", "Burger");
dlist.add(test2, "test2success");// doesn't match first example
Assert.AreEqual("No", dlist.predict(ds.getExample(0)));
}
public void testDecisionListMerge()
{
DecisionList dlist1 = new DecisionList("Yes", "No");
DecisionList dlist2 = new DecisionList("Yes", "No");
DataSet ds = DataSetFactory.getRestaurantDataSet();
tvn.cosine.ai.learning.inductive.DecisionListTest test1 = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test1.add("type", "Thai"); // doesn't match first example
dlist1.add(test1, "test1success");
tvn.cosine.ai.learning.inductive.DecisionListTest test2 = new tvn.cosine.ai.learning.inductive.DecisionListTest();
test2.add("type", "French");
dlist2.add(test2, "test2success");// matches first example
DecisionList dlist3 = dlist1.mergeWith(dlist2);
Assert.AreEqual("test2success", dlist3.predict(ds.getExample(0)));
}
public virtual ICollection <DecisionListTest> createDLTestsWithAttributeCount(DataSet ds, int i)
{
if (i != 1)
{
throw new RuntimeException("For now DLTests with only 1 attribute can be craeted , not" + i);
}
ICollection <string> nonTargetAttributes = ds.getNonTargetAttributes();
ICollection <DecisionListTest> tests = CollectionFactory.CreateQueue <DecisionListTest>();
foreach (string ntAttribute in nonTargetAttributes)
{
ICollection <string> ntaValues = ds.getPossibleAttributeValues(ntAttribute);
foreach (string ntaValue in ntaValues)
{
DecisionListTest test = new DecisionListTest();
test.add(ntAttribute, ntaValue);
tests.Add(test);
}
}
return(tests);
}