uk.ac.starlink.ttools.filter

Class UnivariateStats

java.lang.Object

uk.ac.starlink.ttools.filter.UnivariateStats

public abstract class UnivariateStats
extends Object

Calculates univariate statistics for a variable. Feed data to an instance of this object by repeatedly calling acceptDatum(java.lang.Object, long) and then call the various accessor methods to get accumulated values.

Since:

27 Apr 2006

Author:

Mark Taylor

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static interface	UnivariateStats.ArrayStats Aggregates statistics acquired from a column whose values are fixed-length numeric arrays.

Field Summary

Fields

Modifier and Type	Field and Description
static int	MAX_CARDINALITY Maximum value for cardinality counters.

Constructor Summary

Constructors

Constructor and Description
UnivariateStats()

Method Summary

Modifier and Type	Method and Description
abstract void	acceptDatum(Object value, long irow) Submits a single value to the statistics accumulator.
abstract void	addStats(UnivariateStats other) Adds the accumulated content of a second UnivariateStats object to this one.
static UnivariateStats	createStats(Class<?> clazz, java.util.function.Supplier<Quantiler> qSupplier, boolean doCard) Factory method to construct an instance of this class for accumulating particular types of values.
abstract UnivariateStats.ArrayStats	getArrayStats() Returns an object containing statistics applicable to numeric-array-valued columns.
abstract int	getCardinality() Returns the number of distinct non-null values submitted, if known.
abstract long	getCount() Returns the number of good (non-null) values accumulated.
abstract Comparable<?>	getMaximum() Returns the maximum value submitted, if applicable.
abstract long	getMaxPos() Returns the sequence number of the maximum value submitted.
abstract Comparable<?>	getMinimum() Returns the numeric minimum value submitted, if applicable.
abstract long	getMinPos() Returns the sequence number of the minimum value submitted.
abstract Quantiler	getQuantiler() Returns a quantiler ready to provide quantile values, or null if quantiles were not gathered.
abstract double	getSum() Returns the numeric sum of values accumulated.
abstract double	getSum2() Returns the sum of squares of values accumulated.
abstract double	getSum3() Returns the sum of cubes of values accumulated.
abstract double	getSum4() Returns the sum of fourth powers of values accumulated.

Methods inherited from class java.lang.Object

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

Field Detail

MAX_CARDINALITY

public static final int MAX_CARDINALITY

Maximum value for cardinality counters.

See Also:

Constant Field Values

Constructor Detail

UnivariateStats

public UnivariateStats()

Method Detail

acceptDatum

public abstract void acceptDatum(Object value,
                                 long irow)

Submits a single value to the statistics accumulator. The submitted value should be of a type compatible with the class type of this Stats object.

Parameters:

value - value object

irow - row index of input value

addStats

public abstract void addStats(UnivariateStats other)

Adds the accumulated content of a second UnivariateStats object to this one.

Parameters:

other - compatible UnivariateStats object

getCount

public abstract long getCount()

Returns the number of good (non-null) values accumulated.

Returns:

good value count

getSum

public abstract double getSum()

Returns the numeric sum of values accumulated.

Returns:

sum of values

getSum2

public abstract double getSum2()

Returns the sum of squares of values accumulated.

Returns:

sum of squared values

getSum3

public abstract double getSum3()

Returns the sum of cubes of values accumulated.

Returns:

sum of cubed values

getSum4

public abstract double getSum4()

Returns the sum of fourth powers of values accumulated.

Returns:

sum of fourth powers

getMinimum

public abstract Comparable<?> getMinimum()

Returns the numeric minimum value submitted, if applicable.

Returns:

minimum

getMaximum

public abstract Comparable<?> getMaximum()

Returns the maximum value submitted, if applicable.

Returns:

maximum

getMinPos

public abstract long getMinPos()

Returns the sequence number of the minimum value submitted. Returns -1 if there is no minimum or if the sequence number is not known.

Returns:

row index of minimum, or -1

getMaxPos

public abstract long getMaxPos()

Returns the sequence number of the maximum value submitted. Returns -1 if there is no maximum or if the sequence number is not known.

Returns:

row index of maximum, or -1

getCardinality

public abstract int getCardinality()

Returns the number of distinct non-null values submitted, if known. If the count was not collected, or if there were too many different values to count, -1 is returned.

Returns:

number of distinct non-null values, or -1

getQuantiler

public abstract Quantiler getQuantiler()

Returns a quantiler ready to provide quantile values, or null if quantiles were not gathered. If a non-null quantiler is returned, the Quantiler.ready() value will have been called on it.

Returns:

ready quantiler, or null

getArrayStats

public abstract UnivariateStats.ArrayStats getArrayStats()

Returns an object containing statistics applicable to numeric-array-valued columns.

Returns:

array stats object, or null

createStats

public static UnivariateStats createStats(Class<?> clazz,
                                          java.util.function.Supplier<Quantiler> qSupplier,
                                          boolean doCard)

Factory method to construct an instance of this class for accumulating particular types of values.

Parameters:

clazz - class of which all submitted values will be instances of (if they're not null)

qSupplier - supplier for an object that can calculate quantiles, or null if quantiles are not required

doCard - true if an attempt is to be made to count distinct values

Returns:

stats accumulator