public static void testManualDiscretize() {
Node x = new ContinuousVariable("X");
List<Node> nodes = Collections.singletonList(x);
DataSet data = new ColtDataSet(9, nodes);
data.setDouble(0, 0, 13.0);
data.setDouble(1, 0, 1.2);
data.setDouble(2, 0, 2.2);
data.setDouble(3, 0, 4.5);
data.setDouble(4, 0, 12.005);
data.setDouble(5, 0, 5.5);
data.setDouble(6, 0, 10.1);
data.setDouble(7, 0, 7.5);
data.setDouble(8, 0, 3.4);
System.out.println(data);
Discretizer discretizer = new Discretizer(data);
discretizer.setVariablesCopied(true);
discretizer.equalCounts(x, 3);
DataSet discretized = discretizer.discretize();
System.out.println(discretized);
assertEquals(discretized.getInt(0, 0), 2);
assertEquals(discretized.getInt(1, 0), 0);
assertEquals(discretized.getInt(2, 0), 0);
assertEquals(discretized.getInt(3, 0), 1);
assertEquals(discretized.getInt(4, 0), 2);
assertEquals(discretized.getInt(5, 0), 1);
assertEquals(discretized.getInt(6, 0), 2);
assertEquals(discretized.getInt(7, 0), 1);
assertEquals(discretized.getInt(8, 0), 0);
}
// Causes a package cycle.
public void testManualDiscretize2() {
Graph graph = new Dag(GraphUtils.randomGraph(5, 0, 5, 3, 3, 3, false));
SemPm pm = new SemPm(graph);
SemIm im = new SemIm(pm);
DataSet data = im.simulateData(100, false);
List<Node> nodes = data.getVariables();
Discretizer discretizer = new Discretizer(data);
// discretizer.setVariablesCopied(true);
discretizer.equalCounts(nodes.get(0), 3);
discretizer.equalIntervals(nodes.get(1), 2);
discretizer.equalCounts(nodes.get(2), 5);
discretizer.equalIntervals(nodes.get(3), 8);
discretizer.equalCounts(nodes.get(4), 4);
DataSet discretized = discretizer.discretize();
System.out.println(discretized);
assertEquals(2, maxInColumn(discretized, 0));
assertEquals(1, maxInColumn(discretized, 1));
assertEquals(4, maxInColumn(discretized, 2));
assertEquals(7, maxInColumn(discretized, 3));
assertEquals(3, maxInColumn(discretized, 4));
}
public static void testBreakpointCalculation() {
double[] data = {13, 1.2, 2.2, 4.5, 12.005, 5.5, 10.1, 7.5, 3.4};
double[] breakpoints = Discretizer.getEqualFrequencyBreakPoints(data, 3);
assertTrue(breakpoints.length == 2);
assertEquals(4.5, breakpoints[0]);
assertEquals(10.1, breakpoints[1]);
Discretizer.Discretization dis =
Discretizer.discretize(data, breakpoints, "after", Arrays.asList("0", "1", "2"));
System.out.println(dis);
breakpoints = Discretizer.getEqualFrequencyBreakPoints(data, 4);
assertTrue(breakpoints.length == 3);
assertEquals(3.4, breakpoints[0]);
assertEquals(5.5, breakpoints[1]);
assertEquals(10.1, breakpoints[2]);
}
public void testContinuous() {
final double[] data = {1, 2, 2.5, 3, 4, 5};
double[] cutoffs = new double[] {2.5, 3.2};
List<String> categories = Arrays.asList("lo", "med", "hi");
Discretizer.Discretization discretization =
Discretizer.discretize(data, cutoffs, "after", categories);
System.out.println(discretization);
List<String> discretizedCategories = discretization.getVariable().getCategories();
int[] discretizedData = discretization.getData();
assertEquals("lo", discretizedCategories.get(discretizedData[0]));
assertEquals("lo", discretizedCategories.get(discretizedData[1]));
assertEquals("med", discretizedCategories.get(discretizedData[2]));
assertEquals("med", discretizedCategories.get(discretizedData[3]));
assertEquals("hi", discretizedCategories.get(discretizedData[4]));
assertEquals("hi", discretizedCategories.get(discretizedData[5]));
}
public void testManualDiscretize3() {
Graph graph = new Dag(GraphUtils.randomGraph(5, 0, 5, 3, 3, 3, false));
SemPm pm = new SemPm(graph);
SemIm im = new SemIm(pm);
DataSet data = im.simulateData(100, false);
List<Node> nodes = data.getVariables();
Discretizer discretizer = new Discretizer(data);
discretizer.setVariablesCopied(true);
discretizer.setVariablesCopied(true);
discretizer.equalCounts(nodes.get(0), 3);
DataSet discretized = discretizer.discretize();
System.out.println(discretized);
assertTrue(discretized.getVariable(0) instanceof DiscreteVariable);
assertTrue(discretized.getVariable(1) instanceof ContinuousVariable);
assertTrue(discretized.getVariable(2) instanceof ContinuousVariable);
assertTrue(discretized.getVariable(3) instanceof ContinuousVariable);
assertTrue(discretized.getVariable(4) instanceof ContinuousVariable);
}