/** Orients according to background knowledge */
private void fciOrientbk(IKnowledge bk, Graph graph, List<Node> variables) {
logger.log("info", "Starting BK Orientation.");
for (Iterator<KnowledgeEdge> it = bk.forbiddenEdgesIterator(); it.hasNext(); ) {
KnowledgeEdge edge = it.next();
// match strings to variables in the graph.
Node from = SearchGraphUtils.translate(edge.getFrom(), variables);
Node to = SearchGraphUtils.translate(edge.getTo(), variables);
if (from == null || to == null) {
continue;
}
if (graph.getEdge(from, to) == null) {
continue;
}
// Orient to*->from
graph.setEndpoint(to, from, Endpoint.ARROW);
graph.setEndpoint(from, to, Endpoint.CIRCLE);
changeFlag = true;
logger.log(
"knowledgeOrientation",
SearchLogUtils.edgeOrientedMsg("Knowledge", graph.getEdge(from, to)));
}
for (Iterator<KnowledgeEdge> it = bk.requiredEdgesIterator(); it.hasNext(); ) {
KnowledgeEdge edge = it.next();
// match strings to variables in this graph
Node from = SearchGraphUtils.translate(edge.getFrom(), variables);
Node to = SearchGraphUtils.translate(edge.getTo(), variables);
if (from == null || to == null) {
continue;
}
if (graph.getEdge(from, to) == null) {
continue;
}
graph.setEndpoint(to, from, Endpoint.TAIL);
graph.setEndpoint(from, to, Endpoint.ARROW);
changeFlag = true;
logger.log(
"knowledgeOrientation",
SearchLogUtils.edgeOrientedMsg("Knowledge", graph.getEdge(from, to)));
}
logger.log("info", "Finishing BK Orientation.");
}
/** Returns the pattern to which the given DAG belongs. */
public static Graph patternFromDag(Graph dag) {
Graph graph = new EdgeListGraph(dag);
SearchGraphUtils.basicPattern(graph);
MeekRules rules = new MeekRules();
rules.orientImplied(graph);
return graph;
}
public void layoutByKnowledge() {
GraphWorkbench resultWorkbench = getWorkbench();
Graph graph = resultWorkbench.getGraph();
IKnowledge knowledge = getAlgorithmRunner().getParams().getKnowledge();
SearchGraphUtils.arrangeByKnowledgeTiers(graph, knowledge);
// resultWorkbench.setGraph(graph);
}
/**
* Completes a pattern that was modified by an insertion/deletion operator Based on the algorithm
* described on Appendix C of (Chickering, 2002).
*/
private void rebuildPattern(Graph graph) {
SearchGraphUtils.basicPattern(graph, false);
addRequiredEdges(graph);
meekOrient(graph, getKnowledge());
if (TetradLogger.getInstance().isEventActive("rebuiltPatterns")) {
TetradLogger.getInstance().log("rebuiltPatterns", "Rebuilt pattern = " + graph);
}
}
protected void doDefaultArrangement(Graph resultGraph) {
if (getLatestWorkbenchGraph() != null) { // (alreadyLaidOut) {
GraphUtils.arrangeBySourceGraph(resultGraph, getLatestWorkbenchGraph());
} else if (getKnowledge().isDefaultToKnowledgeLayout()) {
SearchGraphUtils.arrangeByKnowledgeTiers(resultGraph, getKnowledge());
// alreadyLaidOut = true;
} else {
GraphUtils.circleLayout(resultGraph, 200, 200, 150);
// alreadyLaidOut = true;
}
}
private Graph pickDag(Graph graph) {
SearchGraphUtils.basicPattern(graph, false);
addRequiredEdges(graph);
boolean containsUndirected;
do {
containsUndirected = false;
for (Edge edge : graph.getEdges()) {
if (Edges.isUndirectedEdge(edge)) {
containsUndirected = true;
graph.removeEdge(edge);
Edge _edge = Edges.directedEdge(edge.getNode1(), edge.getNode2());
graph.addEdge(_edge);
}
}
meekOrient(graph, getKnowledge());
} while (containsUndirected);
return graph;
}
public void test1() {
for (int r = 0; r < 1; r++) {
Graph mim = DataGraphUtils.randomSingleFactorModel(5, 5, 6, 0, 0, 0);
Graph mimStructure = structure(mim);
SemImInitializationParams params = new SemImInitializationParams();
params.setCoefRange(.5, 1.5);
SemPm pm = new SemPm(mim);
SemIm im = new SemIm(pm, params);
DataSet data = im.simulateData(300, false);
String algorithm = "FOFC";
Graph searchGraph;
List<List<Node>> partition;
if (algorithm.equals("FOFC")) {
FindOneFactorClusters fofc =
new FindOneFactorClusters(data, TestType.TETRAD_WISHART, 0.001);
searchGraph = fofc.search();
partition = fofc.getClusters();
} else if (algorithm.equals("BPC")) {
TestType testType = TestType.TETRAD_WISHART;
TestType purifyType = TestType.TETRAD_BASED2;
BuildPureClusters bpc = new BuildPureClusters(data, 0.001, testType, purifyType);
searchGraph = bpc.search();
partition = MimUtils.convertToClusters2(searchGraph);
} else {
throw new IllegalStateException();
}
List<String> latentVarList = reidentifyVariables(mim, data, partition, 2);
System.out.println(partition);
System.out.println(latentVarList);
System.out.println("True\n" + mimStructure);
Graph mimbuildStructure;
for (int mimbuildMethod : new int[] {3, 4}) {
if (mimbuildMethod == 1) {
System.out.println("Mimbuild 1\n");
Clusters measurements = ClusterUtils.mimClusters(searchGraph);
IndTestMimBuild test = new IndTestMimBuild(data, 0.001, measurements);
MimBuild mimbuild = new MimBuild(test, new Knowledge2());
Graph full = mimbuild.search();
full = changeLatentNames(full, measurements, latentVarList);
mimbuildStructure = structure(full);
System.out.println(
"SHD = "
+ SearchGraphUtils.structuralHammingDistance(mimStructure, mimbuildStructure));
System.out.println("Estimated\n" + mimbuildStructure);
System.out.println();
}
// else if (mimbuildMethod == 2) {
// System.out.println("Mimbuild 2\n");
// Mimbuild2 mimbuild = new Mimbuild2();
// mimbuild.setAlpha(0.001);
// mimbuildStructure = mimbuild.search(partition, latentVarList, data);
// TetradMatrix latentcov = mimbuild.getLatentsCov();
// List<String> latentnames = mimbuild.getLatentNames();
// System.out.println("\nCovariance over the latents");
// System.out.println(MatrixUtils.toStringSquare(latentcov.toArray(),
// latentnames));
// System.out.println("Estimated\n" + mimbuildStructure);
// System.out.println("SHD = " +
// SearchGraphUtils.structuralHammingDistance(mimStructure, mimbuildStructure));
// System.out.println();
// }
else if (mimbuildMethod == 3) {
System.out.println("Mimbuild 3\n");
Mimbuild2 mimbuild = new Mimbuild2();
mimbuild.setAlpha(0.001);
mimbuild.setMinClusterSize(3);
mimbuildStructure = mimbuild.search(partition, latentVarList, new CovarianceMatrix(data));
ICovarianceMatrix latentcov = mimbuild.getLatentsCov();
System.out.println("\nCovariance over the latents");
System.out.println(latentcov);
System.out.println("Estimated\n" + mimbuildStructure);
System.out.println(
"SHD = "
+ SearchGraphUtils.structuralHammingDistance(mimStructure, mimbuildStructure));
System.out.println();
} else if (mimbuildMethod == 4) {
System.out.println("Mimbuild Trek\n");
MimbuildTrek mimbuild = new MimbuildTrek();
mimbuild.setAlpha(0.1);
mimbuild.setMinClusterSize(3);
mimbuildStructure = mimbuild.search(partition, latentVarList, new CovarianceMatrix(data));
ICovarianceMatrix latentcov = mimbuild.getLatentsCov();
System.out.println("\nCovariance over the latents");
System.out.println(latentcov);
System.out.println("Estimated\n" + mimbuildStructure);
System.out.println(
"SHD = "
+ SearchGraphUtils.structuralHammingDistance(mimStructure, mimbuildStructure));
System.out.println();
} else {
throw new IllegalStateException();
}
}
}
}
public void rtest4() {
System.out.println("SHD\tP");
// System.out.println("MB1\tMB2\tMB3\tMB4\tMB5\tMB6");
Graph mim = DataGraphUtils.randomSingleFactorModel(5, 5, 8, 0, 0, 0);
Graph mimStructure = structure(mim);
SemPm pm = new SemPm(mim);
SemImInitializationParams params = new SemImInitializationParams();
params.setCoefRange(0.5, 1.5);
NumberFormat nf = new DecimalFormat("0.0000");
int totalError = 0;
int errorCount = 0;
int maxScore = 0;
int maxNumMeasures = 0;
double maxP = 0.0;
for (int r = 0; r < 1; r++) {
SemIm im = new SemIm(pm, params);
DataSet data = im.simulateData(1000, false);
mim = GraphUtils.replaceNodes(mim, data.getVariables());
List<List<Node>> trueClusters = MimUtils.convertToClusters2(mim);
CovarianceMatrix _cov = new CovarianceMatrix(data);
ICovarianceMatrix cov = DataUtils.reorderColumns(_cov);
String algorithm = "FOFC";
Graph searchGraph;
List<List<Node>> partition;
if (algorithm.equals("FOFC")) {
FindOneFactorClusters fofc =
new FindOneFactorClusters(cov, TestType.TETRAD_WISHART, 0.001f);
searchGraph = fofc.search();
searchGraph = GraphUtils.replaceNodes(searchGraph, data.getVariables());
partition = MimUtils.convertToClusters2(searchGraph);
} else if (algorithm.equals("BPC")) {
TestType testType = TestType.TETRAD_WISHART;
TestType purifyType = TestType.TETRAD_BASED2;
BuildPureClusters bpc = new BuildPureClusters(data, 0.001, testType, purifyType);
searchGraph = bpc.search();
partition = MimUtils.convertToClusters2(searchGraph);
} else {
throw new IllegalStateException();
}
mimStructure = GraphUtils.replaceNodes(mimStructure, data.getVariables());
List<String> latentVarList = reidentifyVariables(mim, data, partition, 2);
Graph mimbuildStructure;
Mimbuild2 mimbuild = new Mimbuild2();
mimbuild.setAlpha(0.001);
mimbuild.setMinClusterSize(3);
try {
mimbuildStructure = mimbuild.search(partition, latentVarList, cov);
} catch (Exception e) {
e.printStackTrace();
continue;
}
mimbuildStructure = GraphUtils.replaceNodes(mimbuildStructure, data.getVariables());
mimbuildStructure = condense(mimStructure, mimbuildStructure);
// Graph mimSubgraph = new EdgeListGraph(mimStructure);
//
// for (Node node : mimSubgraph.getNodes()) {
// if (!mimStructure.getNodes().contains(node)) {
// mimSubgraph.removeNode(node);
// }
// }
int shd = SearchGraphUtils.structuralHammingDistance(mimStructure, mimbuildStructure);
boolean impureCluster = containsImpureCluster(partition, trueClusters);
double pValue = mimbuild.getpValue();
boolean pBelow05 = pValue < 0.05;
boolean numClustersGreaterThan5 = partition.size() != 5;
boolean error = false;
// boolean condition = impureCluster || numClustersGreaterThan5 || pBelow05;
// boolean condition = numClustersGreaterThan5 || pBelow05;
boolean condition = numClustered(partition) == 40;
if (!condition && (shd > 5)) {
error = true;
}
if (!condition) {
totalError += shd;
errorCount++;
}
// if (numClustered(partition) > maxNumMeasures) {
// maxNumMeasures = numClustered(partition);
// maxP = pValue;
// maxScore = shd;
// System.out.println("maxNumMeasures = " + maxNumMeasures);
// System.out.println("maxScore = " + maxScore);
// System.out.println("maxP = " + maxP);
// System.out.println("clusters = " + clusterSizes(partition, trueClusters));
// }
// else
if (pValue > maxP) {
maxScore = shd;
maxP = mimbuild.getpValue();
maxNumMeasures = numClustered(partition);
System.out.println("maxNumMeasures = " + maxNumMeasures);
System.out.println("maxScore = " + maxScore);
System.out.println("maxP = " + maxP);
System.out.println("clusters = " + clusterSizes(partition, trueClusters));
}
System.out.print(
shd
+ "\t"
+ nf.format(pValue)
+ " "
// + (error ? 1 : 0) + " "
// + (pBelow05 ? "P < 0.05 " : "")
// + (impureCluster ? "Impure cluster " : "")
// + (numClustersGreaterThan5 ? "# Clusters != 5 " : "")
// + clusterSizes(partition, trueClusters)
+ numClustered(partition));
System.out.println();
}
System.out.println("\nAvg SHD for not-flagged cases = " + (totalError / (double) errorCount));
System.out.println("maxNumMeasures = " + maxNumMeasures);
System.out.println("maxScore = " + maxScore);
System.out.println("maxP = " + maxP);
}
/**
* Runs PC starting with a commplete graph over the given list of nodes, using the given
* independence test and knowledge and returns the resultant graph. The returned graph will be a
* pattern if the independence information is consistent with the hypothesis that there are no
* latent common causes. It may, however, contain cycles or bidirected edges if this assumption is
* not born out, either due to the actual presence of latent common causes, or due to statistical
* errors in conditional independence judgments.
*
* <p>All of the given nodes must be in the domain of the given conditional independence test.
*/
public Graph search(List<Node> nodes) {
this.logger.log("info", "Starting PC algorithm");
this.logger.log("info", "Independence test = " + getIndependenceTest() + ".");
if (trueDag != null) {
this.dsep = new IndTestDSep(trueDag);
}
long startTime = System.currentTimeMillis();
if (getIndependenceTest() == null) {
throw new NullPointerException();
}
List<Node> allNodes = getIndependenceTest().getVariables();
if (!allNodes.containsAll(nodes)) {
throw new IllegalArgumentException(
"All of the given nodes must " + "be in the domain of the independence test provided.");
}
IFas fas = new Fas2(getIndependenceTest());
fas.setInitialGraph(initialGraph);
fas.setKnowledge(getKnowledge());
fas.setDepth(getDepth());
fas.setVerbose(verbose);
graph = fas.search();
SearchGraphUtils.pcOrientbk(knowledge, graph, nodes);
// independenceTest = new ProbabilisticMAPIndependence((DataSet)
// independenceTest.getData());
SepsetsMaxPValue sepsetProducer =
new SepsetsMaxPValue(graph, independenceTest, null, getDepth());
sepsetProducer.setDsep(dsep);
addColliders(graph, sepsetProducer, knowledge);
MeekRules rules = new MeekRules();
rules.setKnowledge(knowledge);
rules.orientImplied(graph);
// Graph pattern = new EdgeListGraphSingleConnections(graph);
//
// for (Node x : getNodes()) {
// for (Node y : getNodes()) {
// if (x == y) continue;
//
// if (!localMarkovIndep(x, y, pattern, independenceTest)) {
// graph.addUndirectedEdge(x, y);
// }
// }
// }
//
// fas = new FasStableConcurrent(getIndependenceTest());
// fas.setInitialGraph(new EdgeListGraphSingleConnections(graph));
// fas.setKnowledge(getKnowledge());
// fas.setDepth(getDepth());
// fas.setVerbose(verbose);
// graph = fas.search();
//
// sepsetProducer = new SepsetsMaxPValue(graph, independenceTest, null, getDepth());
//
// addColliders(graph, sepsetProducer, knowledge);
//
// rules = new MeekRules();
// rules.setKnowledge(knowledge);
// rules.orientImplied(graph);
this.logger.log("graph", "\nReturning this graph: " + graph);
this.elapsedTime = System.currentTimeMillis() - startTime;
this.logger.log("info", "Elapsed time = " + (elapsedTime) / 1000. + " s");
this.logger.log("info", "Finishing PC Algorithm.");
this.logger.flush();
return graph;
}