Package jslx.statistics

Class Bootstrap

  • All Implemented Interfaces:
    GetNameIfc, IdentityIfc, GetRandomNumberStreamIfc, RNStreamControlIfc, SetRandomNumberStreamIfc
    public class Bootstrap
extends java.lang.Object
implements IdentityIfc, RNStreamControlIfc, SetRandomNumberStreamIfc, GetRandomNumberStreamIfc
    A class to do statistical bootstrapping. The calculations occur via the method generateSamples(). Until generateSamples() is called the results are meaningless. It is possible to save the individual bootstrap samples from which the bootstrap samples can be retrieved. Recognize that this could be a lot of data. The class implements three classic bootstrap confidence intervals normal, basic, and percentile. To estimate the quantiles it uses algorithm 8 from Hyndman, R. J. and Fan, Y. (1996) Sample quantiles in statistical packages, American Statistician 50, 361–365 as the default. This can be changed by the user.

    • Field Detail

      • myId

        protected final int myId
        The id of this object

      • myName

        protected final java.lang.String myName

      • myOriginalPop

        protected final DPopulation myOriginalPop

      • myAcrossBSStat

        protected final Statistic myAcrossBSStat

      • myBSStatList

        protected final java.util.List<Statistic> myBSStatList

      • myOriginalPopStat

        protected final Statistic myOriginalPopStat

      • myPercentileCalc

        protected final org.apache.commons.math3.stat.descriptive.rank.Percentile myPercentileCalc

      • myOrginalData

        protected final double[] myOrginalData

      • myNumBSSamples

        protected int myNumBSSamples

      • myOrgEstimate

        protected double myOrgEstimate

      • myQuantileType

        protected org.apache.commons.math3.stat.descriptive.rank.Percentile.EstimationType myQuantileType

      • myDefaultLevel

        protected double myDefaultLevel

    • Constructor Detail

      • Bootstrap

        public Bootstrap​(double[] originalData)
        Creates a Bootstrap instance with name Bootstrap:getId()
        Parameters:
        originalData - the original data

      • Bootstrap

        public Bootstrap​(java.lang.String name,
                 double[] originalData)
        Parameters:
        name - the name of bootstrap instance
        originalData - the original data

    • Method Detail

      • create

        public static Bootstrap create​(int sampleSize,
                               SampleIfc sampler)
        Parameters:
        sampleSize - the size of the original generate
        sampler - something to generate the original generate of the provided size
        Returns:
        an instance of Bootstrap based on the generate

      • create

        public static Bootstrap create​(java.lang.String name,
                               int sampleSize,
                               SampleIfc sampler)
        Parameters:
        name - the name of bootstrap instance
        sampleSize - the size of the original generate, must be greater than 1
        sampler - something to generate the original generate of the provided size
        Returns:
        an instance of Bootstrap based on the generate

      • generateSamples

        public final void generateSamples​(int numBootstrapSamples)
        The individual bootstrapped samples are not saved. The estimator is EstimatorIfc.Average()
        Parameters:
        numBootstrapSamples - the number of bootstrap samples to generate, must be greater than 1

      • generateSamples

        public final void generateSamples​(int numBootstrapSamples,
                                  boolean saveBootstrapSamples)
        The estimator is EstimatorIfc.Average()
        Parameters:
        numBootstrapSamples - the number of bootstrap samples to generate
        saveBootstrapSamples - indicates that the statistics and data of each bootstrap generate should be saved

      • generateSamples

        public final void generateSamples​(int numBootstrapSamples,
                                  EstimatorIfc estimator)
        The individual bootstrapped samples are not saved.
        Parameters:
        numBootstrapSamples - the number of bootstrap samples to generate
        estimator - a function of the data

      • generateSamples

        public void generateSamples​(int numBootstrapSamples,
                            EstimatorIfc estimator,
                            boolean saveBootstrapSamples)
        This method changes the underlying state of the Bootstrap instance by performing the bootstrap sampling.
        Parameters:
        numBootstrapSamples - the number of bootstrap samples to generate
        estimator - a function of the data
        saveBootstrapSamples - indicates that the statistics and data of each bootstrap generate should be saved

      • getId

        public final int getId()
        Specified by:
        getId in interface IdentityIfc
        Returns:
        the identity is unique to this execution/construction

      • getName

        public final java.lang.String getName()
        Specified by:
        getName in interface GetNameIfc
        Returns:
        the name of the bootstrap

      • resetStartStream

        public void resetStartStream()
        The resetStartStream method will position the RNG at the beginning of its stream. This is the same location in the stream as assigned when the RNG was created and initialized.
        Specified by:
        resetStartStream in interface RNStreamControlIfc

      • resetStartSubstream

        public void resetStartSubstream()
        Resets the position of the RNG at the start of the current substream
        Specified by:
        resetStartSubstream in interface RNStreamControlIfc

      • setAntitheticOption

        public void setAntitheticOption​(boolean flag)
        Tells the stream to start producing antithetic variates
        Specified by:
        setAntitheticOption in interface RNStreamControlIfc
        Parameters:
        flag - true means that it produces antithetic variates.

      • getDefaultCILevel

        public final double getDefaultCILevel()
        Returns:
        the default confidence interval level

      • setDefaultCILevel

        public final void setDefaultCILevel​(double level)
        Parameters:
        level - the level to set must be (0,1)

      • setDefaultQuantileEstimationType

        public final void setDefaultQuantileEstimationType​(org.apache.commons.math3.stat.descriptive.rank.Percentile.EstimationType type)
        Parameters:
        type - the type to set, must not be null.

      • getNumBootstrapSamples

        public final int getNumBootstrapSamples()
        Returns:
        the number of requested bootstrap samples

      • getOriginalData

        public final double[] getOriginalData()
        Returns:
        a copy of the original data

      • getStatisticForEachBootstrapSample

        public final java.util.List<Statistic> getStatisticForEachBootstrapSample()
        Each returned statistic has the data saved. The list itself is unmodifiable. The underlying statistic objects can be modified, but have no effect on the bootstrap generate statistics. The statistical values will be changed the next time generateSamples() is executed. Users are advised to copy the statistics in the list (via Statistic newInstance()) before executing generateSamples if persistence is required. If the save bootstrap data option was not turned on during the sampling then the list returned is empty.
        Returns:
        a list of size getNumBootstrapSamples() holding statistics and data from every bootstrap generate

      • getDataForEachBootstrapSample

        public final java.util.List<double[]> getDataForEachBootstrapSample()
        If the save bootstrap data option was not turned on during the sampling then the list returned is empty.
        Returns:
        a list of size getNumBootstrapSamples() holding a copy of the data from every bootstrap generate

      • getEmpiricalRVForEachBootstrapSample

        public final java.util.List<RVariableIfc> getEmpiricalRVForEachBootstrapSample()
        Creates a random variable to represent the data in each bootstrap generate for which the data was saved. By default, the stream for every random variable is the same across the bootstraps to facilitate common random number generation (CRN)
        Returns:
        a list of the random variables

      • getEmpiricalRVForEachBootstrapSample

        public final java.util.List<RVariableIfc> getEmpiricalRVForEachBootstrapSample​(boolean useCRN)
        Creates a random variable to represent the data in each bootstrap generate for which the data was saved.
        Parameters:
        useCRN - , if true the stream for every random variable is the same across the bootstraps to facilitate common random number generation (CRN). If false different streams are used for each created random variable
        Returns:
        a list of the random variables

      • getDataForBootstrapSample

        public final double[] getDataForBootstrapSample​(int b)
        Parameters:
        b - the bootstrap generate number, b = 1, 2, ... to getNumBootstrapSamples()
        Returns:
        the generated values for the bth bootstrap, if no samples are saved then the array returned is of zero length

      • getBootstrapSampleAverages

        public final double[] getBootstrapSampleAverages()
        If the bootstrap samples were saved, this returns the generated averages for each of the samples
        Returns:
        an array of the bootstrap generate averages, will be zero length if no bootstrap samples were saved

      • getBootstrapSampleVariances

        public final double[] getBootstrapSampleVariances()
        If the bootstrap samples were saved, this returns the generated variance for each of the samples
        Returns:
        an array of the bootstrap generated variances, will be zero length if no bootstrap samples were saved

      • getAcrossBootstrapStatistics

        public final Statistic getAcrossBootstrapStatistics()
        Returns:
        a Statistic observed over estimates from the bootstrap samples

      • getAcrossBootstrapAverage

        public final double getAcrossBootstrapAverage()
        Returns:
        the generated average of the estimates from the bootstrap samples

      • getBootstrapEstimates

        public final double[] getBootstrapEstimates()
        Returns:
        the observations of the estimator for each bootstrap generate, may be zero length if no samples have been generated

      • getBootstrapDifferences

        public final double[] getBootstrapDifferences()
        Each element is the bootstrap estimate for sample i minus getOriginalDataEstimate()
        Returns:
        the array of bootstrap differences

      • getStandardizedBootstrapDifferences

        public final double[] getStandardizedBootstrapDifferences()
        Each element is the bootstrap estimate for sample i minus getOriginalDataEstimate() divided by getBootstrapStdErrEstimate()
        Returns:
        the array of bootstrap differences

      • getOriginalDataAverage

        public final double getOriginalDataAverage()
        Returns:
        the generate average for the original data

      • getOriginalDataEstimate

        public final double getOriginalDataEstimate()
        Returns:
        the estimate from the supplied EstimatorIfc based on the original data

      • getBootstrapBiasEstimate

        public final double getBootstrapBiasEstimate()
        This is getAcrossBootstrapAverage() - getOriginalDataEstimate()
        Returns:
        an estimate the bias based on bootstrapping

      • getBootstrapStdErrEstimate

        public final double getBootstrapStdErrEstimate()
        This is the standard deviation of the across bootstrap observations of the estimator for each bootstrap generate
        Returns:
        the standard error of the estimate based on bootstrapping

      • getOriginalDataStatistics

        public final Statistic getOriginalDataStatistics()
        Returns:
        summary statistics for the original data

      • getStdNormalBootstrapCI

        public final Interval getStdNormalBootstrapCI()
        Gets the standard normal based bootstrap confidence interval. Not recommended.
        Returns:
        the confidence interval

      • getStdNormalBootstrapCI

        public final Interval getStdNormalBootstrapCI​(double level)
        Gets the standard normal based bootstrap confidence interval. Not recommended.
        Parameters:
        level - the confidence level, must be between 0 and 1
        Returns:
        the confidence interval

      • getBasicBootstrapCI

        public final Interval getBasicBootstrapCI()
        The "basic" method, but with no bias correction. This is the so called centered percentile method (2θ − Bu , 2θ − Bl ) where θ is the bootstrap estimator and Bu is the 1 - alpha/2 percentile and Bl is the lower (alpha/2) percentile, where level = 1-alpha of the bootstrap replicates.
        Returns:
        the confidence interval

      • getBasicBootstrapCI

        public final Interval getBasicBootstrapCI​(double level)
        The "basic" method, but with no bias correction. This is the so called centered percentile method (2θ − Bu , 2θ − Bl ) where θ is the bootstrap estimator and Bu is the 1 - alpha/2 percentile and Bl is the lower (alpha/2) percentile, where level = 1-alpha of the bootstrap replicates.
        Parameters:
        level - the confidence level, must be between 0 and 1
        Returns:
        the confidence interval

      • getPercentileBootstrapCI

        public final Interval getPercentileBootstrapCI()
        The "percentile" method, but with no bias correction. This is the percentile method (θ − Bl , θ + Bu ) where θ is the bootstrap estimator and Bu is the 1 - alpha/2 percentile and Bl is the lower (alpha/2) percentile, where level = 1-alpha of the bootstrap replicates
        Returns:
        the confidence interval

      • getPercentileBootstrapCI

        public final Interval getPercentileBootstrapCI​(double level)
        The "percentile" method, but with no bias correction. This is the percentile method (Bl , Bu ) where Bu is the 1 - alpha/2 percentile and Bl is the lower (alpha/2) percentile, where level = 1-alpha of the bootstrap replicates
        Parameters:
        level - the confidence level, must be between 0 and 1
        Returns:
        the confidence interval

      • toString

        public java.lang.String toString()
        Overrides:
        toString in class java.lang.Object

      • asString

        public java.lang.String asString()