Package edu.cmu.tetrad.search

Class Ida

java.lang.Object

edu.cmu.tetrad.search.Ida

public class Ida
extends Object

Implements the IDA algorithm. The reference is here:

Maathuis, Marloes H., Markus Kalisch, and Peter Bühlmann. "Estimating high-dimensional intervention effects from observational data." The Annals of Statistics 37.6A (2009): 3133-3164.

The IDA algorithm seeks to give a list of possible parents of a given variable Y and their corresponding lower-bounded effects on Y. It regresses Y on X ∪ S, where X is a possible parent of Y and S is a set of possible parents of X, and reports the regression coefficient. The set of such regressions is then sorted in ascending order to give the total effects of X on Y. The absolute total effects are calculated as the absolute values of the total effects.

Version:

$Id: $Id

Author:

josephramsey

See Also:

• Cstar
• Ida.NodeEffects

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	Ida.NodeEffects	Gives a list of nodes (parents or children) and corresponding minimum effects for the IDA algorithm.

Constructor Summary

Constructors

Constructor	Description
Ida(DataSet dataSet, Graph graph, List<Node> possibleCauses)	Constructor.

Method Summary

Modifier and Type	Method	Description
Map<Node,Double>	calculateMinimumTotalEffectsOnY(Node y)	Returns a map from nodes in V \ {Y} to their minimum effects.
double	distance(LinkedList<Double> effects, double trueEffect)	Returns the distance between the effects and the true effect.
LinkedList<Double>	getAbsTotalEffects(Node x, Node y)	This method calculates the absolute total effects of node x on node y.
LinkedList<Double>	getTotalEffects(Node x, Node y)	Calculates the total effects of node x on node y.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

Ida

public Ida(DataSet dataSet,
 Graph graph,
 List<Node> possibleCauses)

Constructor.

Parameters:

dataSet - The dataset being searched over.

graph - The graph model. Should be a DAG or a CPDAG.

possibleCauses - The possible causes to be considered.

Method Details

distance

public double distance(LinkedList<Double> effects,
 double trueEffect)

Returns the distance between the effects and the true effect.

Parameters:

effects - a LinkedList object

trueEffect - a double

Returns:

This difference.

getTotalEffects

public LinkedList<Double> getTotalEffects(Node x,
 Node y)

Calculates the total effects of node x on node y.

Parameters:

x - The node whose total effects are to be calculated.

y - The node for which the total effects are calculated.

Returns:

A LinkedList of Double values representing the total effects of node x on node y. The LinkedList is sorted in ascending order.

getAbsTotalEffects

public LinkedList<Double> getAbsTotalEffects(Node x,
 Node y)

This method calculates the absolute total effects of node x on node y.

Parameters:

x - The node for which the total effects are calculated.

y - The node whose total effects are obtained.

Returns:

A LinkedList of Double values representing the absolute total effects of node x on node y. The LinkedList is sorted in ascending order.

calculateMinimumTotalEffectsOnY

public Map<Node,Double> calculateMinimumTotalEffectsOnY(Node y)

Returns a map from nodes in V \ {Y} to their minimum effects.

Parameters:

y - The child variable

Returns:

Thia map.