Package edu.cmu.tetrad.search.score

Class GraphScore

java.lang.Object
edu.cmu.tetrad.search.score.GraphScore
All Implemented Interfaces:
Score
public class GraphScore extends Object implements Score

Implements a pscudo-"score" that implmenets implements Chickering and Meek's (2002) locally consistent score criterion. This is not a true score; rather, a -1 is returned in case mseparation holds and a 1 in case mseparation does not hold. This is only meant to be used in the context of FGES, and allows the search to follow its path prescribed by the locally consistent scoring criterion. For a reference to the latter, pleasee this article:

Chickering (2002) "Optimal structure identification with greedy search" Journal of Machine Learning Research.

For further discussion of using m-separation in the GES search, see:

Nandy, P., Hauser, A., & Maathuis, M. H. (2018). High-dimensional consistency in score-based and hybrid structure learning. The Annals of Statistics, 46(6A), 3151-3183.

For more discussion please see:

Shen, X., Zhu, S., Zhang, J., Hu, S., & Chen, Z. (2022, August). Reframed GES with a neural conditional dependence measure. In Uncertainty in Artificial Intelligence (pp. 1782-1791). PMLR.

Author:
josephramsey
See Also:

  • Constructor Details

    • GraphScore

      public GraphScore(Graph dag)
      Constructor
      Parameters:
      dag - A directed acyclic graph.

    • GraphScore

      public GraphScore(IndependenceFacts facts)
      Constructor.
      Parameters:
      facts - A list known independence facts; a lookup will be donw from these facts.
      See Also:

  • Method Details

    • localScore

      public double localScore(int y, int[] z)
      Calculates the sample likelihood and BIC score for y given its z in a simple SEM model.
      Specified by:
      localScore in interface Score
      Parameters:
      y - The node.
      z - The parents.
      Returns:
      this score.

    • localScoreDiff

      public double localScoreDiff(int x, int y, int[] z)
      Returns a "score difference", which amounts to a conditional local scoring criterion results
      Specified by:
      localScoreDiff in interface Score
      Parameters:
      x - A node.
      y - TAhe node.
      z - A set of nodes.
      Returns:
      The "difference".

    • localScoreDiff

      public double localScoreDiff(int x, int y)
      The "unconditional difference."
      Specified by:
      localScoreDiff in interface Score
      Parameters:
      x - A node.
      y - The node.
      Returns:
      This.

    • localScore

      public double localScore(int i, int parent)
      Description copied from interface: Score
      Returns the local score of the graph.
      Specified by:
      localScore in interface Score
      Parameters:
      i - A node.
      parent - A parent.
      Returns:
      The local score.
      Throws:
      javax.help.UnsupportedOperationException - Since the method doesn't make sense here.

    • localScore

      public double localScore(int i)
      Description copied from interface: Score
      Returns the local score of the gien node in the graph.
      Specified by:
      localScore in interface Score
      Parameters:
      i - A node.
      Returns:
      The local score.
      Throws:
      javax.help.UnsupportedOperationException - Since the method doesn't make sense here.

    • isEffectEdge

      public boolean isEffectEdge(double bump)
      Returns a judgment for FGES as to whether a score with the bump is for an effect edge.
      Specified by:
      isEffectEdge in interface Score
      Parameters:
      bump - The bump
      Returns:
      True, if so.
      See Also:

    • getDataSet

      public DataSet getDataSet()
      Throws:
      javax.help.UnsupportedOperationException - Since the method doesn't make sense here.

    • getVariables

      public List<Node> getVariables()
      Returns the list of variables.
      Specified by:
      getVariables in interface Score
      Returns:
      This list.

    • getMaxDegree

      public int getMaxDegree()
      Returns the maximum degree, which is set to 1000.
      Specified by:
      getMaxDegree in interface Score
      Returns:
      1000.

    • getData

      public DataModel getData()
      Throws:
      javax.help.UnsupportedOperationException - Since this "score" does not use data.

    • getSampleSize

      public int getSampleSize()
      Description copied from interface: Score
      The sample size of the data.
      Specified by:
      getSampleSize in interface Score
      Returns:
      This size.
      Throws:
      javax.help.UnsupportedOperationException - Since this score does not use data.

    • getDag

      public Graph getDag()
      Returns a copy of the DAG being searched over.
      Returns:
      This DAG.