Examples with Dag edu.cmu.tetrad.graph.Dag used on opensource projects

Example 16 with Dag

use of edu.cmu.tetrad.graph.Dag in project tetrad by cmu-phil.

the class TestRowSummingUpdater method sampleBayesIm2.

private BayesIm sampleBayesIm2() {
    Node a = new GraphNode("a");
    Node b = new GraphNode("b");
    Node c = new GraphNode("c");
    Dag graph;
    graph = new Dag();
    graph.addNode(a);
    graph.addNode(b);
    graph.addNode(c);
    graph.addDirectedEdge(a, b);
    graph.addDirectedEdge(a, c);
    graph.addDirectedEdge(b, c);
    BayesPm bayesPm = new BayesPm(graph);
    bayesPm.setNumCategories(b, 3);
    BayesIm bayesIm1 = new MlBayesIm(bayesPm);
    bayesIm1.setProbability(0, 0, 0, .3);
    bayesIm1.setProbability(0, 0, 1, .7);
    bayesIm1.setProbability(1, 0, 0, .3);
    bayesIm1.setProbability(1, 0, 1, .4);
    bayesIm1.setProbability(1, 0, 2, .3);
    bayesIm1.setProbability(1, 1, 0, .6);
    bayesIm1.setProbability(1, 1, 1, .1);
    bayesIm1.setProbability(1, 1, 2, .3);
    bayesIm1.setProbability(2, 0, 0, .9);
    bayesIm1.setProbability(2, 0, 1, .1);
    bayesIm1.setProbability(2, 1, 0, .1);
    bayesIm1.setProbability(2, 1, 1, .9);
    bayesIm1.setProbability(2, 2, 0, .5);
    bayesIm1.setProbability(2, 2, 1, .5);
    bayesIm1.setProbability(2, 3, 0, .2);
    bayesIm1.setProbability(2, 3, 1, .8);
    bayesIm1.setProbability(2, 4, 0, .6);
    bayesIm1.setProbability(2, 4, 1, .4);
    bayesIm1.setProbability(2, 5, 0, .7);
    bayesIm1.setProbability(2, 5, 1, .3);
    return bayesIm1;
}

Example 17 with Dag

use of edu.cmu.tetrad.graph.Dag in project tetrad by cmu-phil.

the class TestMatrixUtils method testImpiedCovar.

@Test
public void testImpiedCovar() {
    List<Node> nodes = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        nodes.add(new ContinuousVariable("X" + (i + 1)));
    }
    Graph graph = new Dag(GraphUtils.randomGraph(nodes, 0, 10, 30, 15, 15, false));
    SemPm pm = new SemPm(graph);
    SemIm im = new SemIm(pm);
    TetradMatrix err = im.getErrCovar();
    TetradMatrix coef = im.getEdgeCoef();
    TetradMatrix implied = MatrixUtils.impliedCovar(coef, err);
    assertTrue(MatrixUtils.isPositiveDefinite(implied));
    TetradMatrix corr = MatrixUtils.convertCovToCorr(new TetradMatrix(implied));
    assertTrue(MatrixUtils.isPositiveDefinite(corr));
}

Example 18 with Dag

use of edu.cmu.tetrad.graph.Dag in project tetrad by cmu-phil.

the class EmBayesEstimator method findBayesNetObserved.

private void findBayesNetObserved() {
    Dag dagObs = new Dag(graph);
    for (Node node : nodes) {
        if (node.getNodeType() == NodeType.LATENT) {
            dagObs.removeNode(node);
        }
    }
    bayesPmObs = new BayesPm(dagObs, bayesPm);
// System.out.println("bayesPm after deleting edges involving latents:");
// System.out.println(bayesPmObs);
// Graph g = bayesPmObs.getDag();
// System.out.println(g);
}

Example 19 with Dag

use of edu.cmu.tetrad.graph.Dag in project tetrad by cmu-phil.

the class FactoredBayesStructuralEM method scoreTest.

// public double returnScore() {
// return resultScore;
// }
// 
// public Graph returnGraph() {
// return resultGraph;
// }
// 
// public int returnIterations() {
// return totaliterations;
// }
public void scoreTest() {
    TetradLogger.getInstance().log("details", "scoreTest");
    // System.out.println(bayesPmM0.getGraph());
    BdeMetricCache bdeMetricCache;
    BayesPm bayesPmMn = bayesPmM0;
    EmBayesEstimator emBayesEst = new EmBayesEstimator(bayesPmMn, dataSet);
    emBayesEst.maximization(0.0001);
    Dag dag0 = new Dag(bayesPmMn.getDag());
    // double score = factorScoreMD(dag0, bdeMetricCache, bayesPmMn, bayesImMn);
    // System.out.println("Score of dag0 = " + score);
    Node L1 = dag0.getNode("L1");
    Node X1 = dag0.getNode("X1");
    Dag dag1 = new Dag(dag0);
    dag1.addDirectedEdge(X1, L1);
    // BayesPm bayesPmTest = new BayesPm(dag0);
    // score = factorScoreMD(dag0, bdeMetricCache, bayesPmTest, bayesImMn);
    // System.out.println("Score of dag1 = " + score);
    BayesPm bayesPm0 = new BayesPm(dag0);
    EmBayesEstimator emBayesEst0 = new EmBayesEstimator(bayesPm0, dataSet);
    BayesIm bayesImMn0 = emBayesEst0.maximization(0.0001);
    BayesPm bayesPmTest0 = new BayesPm(dag0);
    TetradLogger.getInstance().log("details", "Observed conts for nodes of L1,X1,X2,X3 (no edges) " + "using the MAP parameters based on that same graph");
    TetradLogger.getInstance().log("details", "Graph of PM:  ");
    TetradLogger.getInstance().log("details", "" + bayesPmTest0.getDag());
    TetradLogger.getInstance().log("details", "Graph of IM:  ");
    TetradLogger.getInstance().log("details", "" + bayesImMn0.getBayesPm().getDag());
    bdeMetricCache = new BdeMetricCache(dataSet, bayesPmTest0);
    List<Node> nodes0 = dag0.getNodes();
    for (Node aNodes0 : nodes0) {
        double[][] counts0 = bdeMetricCache.getObservedCounts(aNodes0, bayesPmTest0, bayesImMn0);
        for (double[] aCounts0 : counts0) {
            for (int j = 0; j < counts0[0].length; j++) {
                System.out.print(" " + aCounts0[j]);
            }
            TetradLogger.getInstance().log("details", "\n");
        }
        TetradLogger.getInstance().log("details", "\n");
    }
    double score0 = factorScoreMD(dag0, bdeMetricCache, bayesPmTest0, bayesImMn0);
    TetradLogger.getInstance().log("details", "Score of L1,X1,X2,X3 (no edges) for itself = " + score0);
    TetradLogger.getInstance().log("details", "===============\n\n");
    TetradLogger.getInstance().log("details", "Score of X1-->L1 for L1,X1,X2,X3 (no edges) = " + score0);
    BayesPm bayesPmTest1 = new BayesPm(dag1);
    TetradLogger.getInstance().log("details", "Observed counts for nodes of X1-->L1 for L1,X1,X2,X3 (no edges)");
    TetradLogger.getInstance().log("details", "Graph of PM :  ");
    TetradLogger.getInstance().log("details", "" + bayesPmTest1.getDag());
    TetradLogger.getInstance().log("details", "Graph of IM:  ");
    TetradLogger.getInstance().log("details", "" + bayesImMn0.getBayesPm().getDag());
    bdeMetricCache = new BdeMetricCache(dataSet, bayesPmTest1);
    List<Node> nodes1 = dag0.getNodes();
    for (Node aNodes1 : nodes1) {
        double[][] counts1 = bdeMetricCache.getObservedCounts(aNodes1, bayesPmTest1, bayesImMn0);
        for (double[] aCounts1 : counts1) {
            for (int j = 0; j < counts1[0].length; j++) {
                TetradLogger.getInstance().log("details", " " + aCounts1[j]);
            }
            TetradLogger.getInstance().log("details", "\n");
        }
        TetradLogger.getInstance().log("details", "\n");
    }
    double score1 = factorScoreMD(dag1, bdeMetricCache, bayesPmTest1, bayesImMn0);
    TetradLogger.getInstance().log("details", "Score of X1-->L1 for L1,X1,X2,X3 (no edges) = " + score1);
}

Example 20 with Dag

use of edu.cmu.tetrad.graph.Dag in project tetrad by cmu-phil.

the class BayesXmlParser method makeBayesPm.

private BayesPm makeBayesPm(List<Node> variables, Element element1) {
    if (!"parents".equals(element1.getQualifiedName())) {
        throw new IllegalArgumentException("Expecting 'parents' element.");
    }
    Dag graph = new Dag();
    for (Node variable : variables) {
        graph.addNode(variable);
    }
    Elements elements = element1.getChildElements();
    for (int i = 0; i < elements.size(); i++) {
        Element e1 = elements.get(i);
        if (!"parentsFor".equals(e1.getQualifiedName())) {
            throw new IllegalArgumentException("Expecting 'parentsFor' element.");
        }
        String varName = e1.getAttributeValue("name");
        Node var = namesToVars.get(varName);
        Elements elements1 = e1.getChildElements();
        for (int j = 0; j < elements1.size(); j++) {
            Element e2 = elements1.get(j);
            if (!"parent".equals(e2.getQualifiedName())) {
                throw new IllegalArgumentException("Expecting 'parent' element.");
            }
            String parentName = e2.getAttributeValue("name");
            Node parent = namesToVars.get(parentName);
            graph.addDirectedEdge(parent, var);
        }
    }
    BayesPm bayesPm = new BayesPm(graph);
    for (Node variable1 : variables) {
        DiscreteVariable graphVariable = (DiscreteVariable) variable1;
        List<String> categories = graphVariable.getCategories();
        bayesPm.setCategories(graphVariable, categories);
    }
    return bayesPm;
}