Package com.imsl.stat

Class SelectionRegression.Statistics

java.lang.Object
com.imsl.stat.SelectionRegression.Statistics
All Implemented Interfaces:
Serializable
Enclosing class:
SelectionRegression
public class SelectionRegression.Statistics extends Object implements Serializable
Statistics contains statistics related to the regression coefficients.
See Also:

  • Method Details

    • getCriterionValues

      public double[] getCriterionValues(int numVariables)
      Returns an array containing the best criterion values for variable subsets of a given size.
      Parameters:
      numVariables - An int which specifies the variable subset size.
      Returns:
      A double array of size less than or equal to maxSaved containing the criterion values.

    • getIndependentVariables

      public int[][] getIndependentVariables(int numVariables)
      Returns the identification numbers for the independent variable subsets of a given size in the same order as the criteria returned by getCriterionValues(int).
      Parameters:
      numVariables - An int which specifies the variable subset size.
      Returns:
      An int matrix of size k by numVariables, where k is less than or equal to maxSaved, containing the identification numbers for the independent variable subsets considered.

    • getCoefficientStatistics

      public double[][] getCoefficientStatistics(int regressionIndex)
      Returns the coefficients statistics for each of the best regressions found for each subset considered.

      The value set by method SelectionRegression.setMaximumBestFound(int) determines the total number of best regressions to find. The number of best regressions is equal to (maxSubset x maxFound), if criterion R_SQUARED_CRITERION is specified or it is equal to maxFound if either MALLOWS_CP_CRITERION or ADJUSTED_R_SQUARED_CRITERION is specified.

      Each row contains statistics related to the regression coefficients of the best models. The regressions are ordered so that the better regressions appear first. The statistic in the columns are as follows (inferences are conditional on the selected model):

      Column Description
      0 variable number
      1 coefficient estimate
      2 estimated standard error of the estimate
      3 t-statistic for the test that the coefficient is 0
      4 p-value for the two-sided t test

      Parameters:
      regressionIndex - An int which specifies the index of the best regression statistics to return. There will be 0 to (maxSubset x maxFound - 1) best regressions if R_SQUARED_CRITERION is specified or 0 to (maxFound - 1) if either MALLOWS_CP_CRITERION or ADJUSTED_R_SQUARED_CRITERION is specified.
      Returns:
      A two-dimensional double array containing the regression statistics.
      See Also: