Package org.apache.datasketches.sampling

Class VarOptItemsSketch<T>

java.lang.Object
org.apache.datasketches.sampling.VarOptItemsSketch<T>
Type Parameters:
T - The type of object held in the sketch.
public final class VarOptItemsSketch<T> extends Object
This sketch provides a variance optimal sample over an input stream of weighted items. The sketch can be used to compute subset sums over predicates, producing estimates with optimal variance for a given sketch size.

Using this sketch with uniformly constant item weights (e.g. 1.0) will produce a standard reservoir sample over the steam.

Author:
Jon Malkin, Kevin Lang

  • Method Summary

    Modifier and Type
    Method
    Description
    SampleSubsetSummary
    estimateSubsetSum(Predicate<T> predicate)
    Computes an estimated subset sum from the entire stream for objects matching a given predicate.
    int
    getK()
    Returns the sketch's value of k, the maximum number of samples stored in the sketch.
    long
    getN()
    Returns the number of items processed from the input stream
    int
    getNumSamples()
    Returns the current number of items in the sketch, which may be smaller than the sketch capacity.
    VarOptItemsSamples<T>
    getSketchSamples()
    Gets a result iterator object.
    static <T> VarOptItemsSketch<T>
    heapify(org.apache.datasketches.memory.Memory srcMem, ArrayOfItemsSerDe<T> serDe)
    Returns a sketch instance of this class from the given srcMem, which must be a Memory representation of this sketch class.
    static <T> VarOptItemsSketch<T>
    newInstance(int k)
    Construct a varopt sampling sketch with up to k samples using the default resize factor (8).
    static <T> VarOptItemsSketch<T>
    newInstance(int k, ResizeFactor rf)
    Construct a varopt sampling sketch with up to k samples using the specified resize factor.
    void
    reset()
    Resets this sketch to the empty state, but retains the original value of k.
    byte[]
    toByteArray(ArrayOfItemsSerDe<? super T> serDe)
    Returns a byte array representation of this sketch.
    byte[]
    toByteArray(ArrayOfItemsSerDe<? super T> serDe, Class<?> clazz)
    Returns a byte array representation of this sketch.
    String
    toString()
    Returns a human-readable summary of the sketch.
    static String
    toString(byte[] byteArr)
    Returns a human readable string of the preamble of a byte array image of a VarOptItemsSketch.
    static String
    toString(org.apache.datasketches.memory.Memory mem)
    Returns a human readable string of the preamble of a Memory image of a VarOptItemsSketch.
    void
    update(T item, double weight)
    Randomly decide whether or not to include an item in the sample set.

    Methods inherited from class java.lang.Object

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

  • Method Details

    • newInstance

      public static <T> VarOptItemsSketch<T> newInstance(int k)
      Construct a varopt sampling sketch with up to k samples using the default resize factor (8).
      Type Parameters:
      T - The type of object held in the sketch.
      Parameters:
      k - Maximum size of sampling. Allocated size may be smaller until sketch fills. Unlike many sketches in this package, this value does not need to be a power of 2.
      Returns:
      A VarOptItemsSketch initialized with maximum size k and resize factor rf.

    • newInstance

      public static <T> VarOptItemsSketch<T> newInstance(int k, ResizeFactor rf)
      Construct a varopt sampling sketch with up to k samples using the specified resize factor.
      Type Parameters:
      T - The type of object held in the sketch.
      Parameters:
      k - Maximum size of sampling. Allocated size may be smaller until sketch fills. Unlike many sketches in this package, this value does not need to be a power of 2. The maximum size is Integer.MAX_VALUE-1.
      rf - See Resize Factor
      Returns:
      A VarOptItemsSketch initialized with maximum size k and resize factor rf.

    • heapify

      public static <T> VarOptItemsSketch<T> heapify(org.apache.datasketches.memory.Memory srcMem, ArrayOfItemsSerDe<T> serDe)
      Returns a sketch instance of this class from the given srcMem, which must be a Memory representation of this sketch class.
      Type Parameters:
      T - The type of item this sketch contains
      Parameters:
      srcMem - a Memory representation of a sketch of this class. See Memory
      serDe - An instance of ArrayOfItemsSerDe
      Returns:
      a sketch instance of this class

    • getK

      public int getK()
      Returns the sketch's value of k, the maximum number of samples stored in the sketch. The current number of items in the sketch may be lower.
      Returns:
      k, the maximum number of samples in the sketch

    • getN

      public long getN()
      Returns the number of items processed from the input stream
      Returns:
      n, the number of stream items the sketch has seen

    • getNumSamples

      public int getNumSamples()
      Returns the current number of items in the sketch, which may be smaller than the sketch capacity.
      Returns:
      the number of items currently in the sketch

    • getSketchSamples

      public VarOptItemsSamples<T> getSketchSamples()
      Gets a result iterator object.
      Returns:
      An object with an iterator over the results

    • update

      public void update(T item, double weight)
      Randomly decide whether or not to include an item in the sample set.
      Parameters:
      item - an item of the set being sampled from
      weight - a strictly positive weight associated with the item

    • reset

      public void reset()
      Resets this sketch to the empty state, but retains the original value of k.

    • toString

      public String toString()
      Returns a human-readable summary of the sketch.
      Overrides:
      toString in class Object
      Returns:
      A string version of the sketch summary

    • toString

      public static String toString(byte[] byteArr)
      Returns a human readable string of the preamble of a byte array image of a VarOptItemsSketch.
      Parameters:
      byteArr - the given byte array
      Returns:
      a human readable string of the preamble of a byte array image of a VarOptItemsSketch.

    • toString

      public static String toString(org.apache.datasketches.memory.Memory mem)
      Returns a human readable string of the preamble of a Memory image of a VarOptItemsSketch.
      Parameters:
      mem - the given Memory
      Returns:
      a human readable string of the preamble of a Memory image of a VarOptItemsSketch.

    • toByteArray

      public byte[] toByteArray(ArrayOfItemsSerDe<? super T> serDe)
      Returns a byte array representation of this sketch. May fail for polymorphic item types.
      Parameters:
      serDe - An instance of ArrayOfItemsSerDe
      Returns:
      a byte array representation of this sketch

    • toByteArray

      public byte[] toByteArray(ArrayOfItemsSerDe<? super T> serDe, Class<?> clazz)
      Returns a byte array representation of this sketch. Copies contents into an array of the specified class for serialization to allow for polymorphic types.
      Parameters:
      serDe - An instance of ArrayOfItemsSerDe
      clazz - The class represented by <T>
      Returns:
      a byte array representation of this sketch

    • estimateSubsetSum

      public SampleSubsetSummary estimateSubsetSum(Predicate<T> predicate)
      Computes an estimated subset sum from the entire stream for objects matching a given predicate. Provides a lower bound, estimate, and upper bound using a target of 2 standard deviations.

      This is technically a heuristic method, and tries to err on the conservative side.

      Parameters:
      predicate - A predicate to use when identifying items.
      Returns:
      A summary object containing the estimate, upper and lower bounds, and the total sketch weight.