mml

Class UPModel

java.lang.Object

la.la.Value

la.la.Function

la.la.Function.Native

mml.UPModel

All Implemented Interfaces:

java.lang.Comparable<Value>

Direct Known Subclasses:

BestOf, ByPdf, Discretes, Graphs, Missing, Mixture, Model.Transform, Multinomial.M.Trials, Multivariate, Permutation, Sequences, UPFunctionModel, UPModel.Transform, UPSeriesModel

public abstract class UPModel
extends Function.Native

UPModel, the abstract class of UnParameterised statistical Models. Fully parameterised Models are UPModel.M and Model; also see Estimator. Statistical parameters are not yet set in a UPModel but it holds any common-knowledge, problem-defining parameters, and returns a Model when it is applied to statistical-parameter(s) sp (or calls sp2Model( msg1, msg2, sp)), and hopefully can also estimate a fully parameterised Model from a given data-set.
Useful examples include Normal, NormalMu, Poisson0, Geometric0, and Multivariate.
Also see UPFunctionModel and UPSeriesModel.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	UPModel.Est Est, an Estimator that uses information from its enclosing UnParameterised Model.
class	UPModel.M A fully parameterised Model, M, nested within an UnParameterised UPModel.
class	UPModel.Transform Transform 'this' UPModel with Function f.

Nested classes/interfaces inherited from class la.la.Function.Native

Function.Native.WithInverse

Nested classes/interfaces inherited from class la.la.Function

Function.Cts2Cts, Function.Cts2Cts2Cts, Function.CtsD2CtsD, Function.HasInverse, Function.Native, Function.Native2, Function.Native3

Nested classes/interfaces inherited from class la.la.Value

Value.Atomic, Value.Bool, Value.Char, Value.Chars, Value.Cts, Value.Defer, Value.Discrete, Value.Enum, Value.Inc_Or, Value.Int, Value.Lambda, Value.List, Value.Maybe, Value.Option, Value.Real, Value.Scannable, Value.Structured, Value.Triv, Value.Tuple

Field Summary

Fields

Modifier and Type	Field and Description
protected Value	dp The problem-defining parameter(s) of 'this' UnParameterised Model; see defnParams().

Fields inherited from class la.la.Value

C, comparator, CR, E, eight, ffalse, five, fiveR, four, fourR, half, negCR, negOne, negOneR, negTenR, nilVal, nine, None, one, oneR, PI, PIby2, point01, point1, point5, point9, quarter, seven, six, ten, tenR, three, threeR, triv, ttrue, two, twoR, zero, zeroR

Constructor Summary

Constructors

Constructor and Description
UPModel(Value dp) Given common-knowledge problem-defining parameter(s), dp, create an UnParameterised Model.

Method Summary

Modifier and Type	Method and Description
Model	apply(Value sp) Given statistical parameters, sp, return a Model, sp2Model(0, 0, sp), i.e., one that is not estimated, having zero part 1 and part 2 message lengths.
Value	defnParams() Return the common-knowledge, problem-defining parameter(s) for the fully parameterised M-Models that 'this' UPModel can and will produce.
Estimator	estimator(Value ps) Given parameter(s), 'ps' (possibly parameters of a prior), return an Estimator for 'this' UPModel.
static void	main(java.lang.String[] argv) Nothing new here; see Test.
abstract Model	sp2Model(double msg1, double msg2, Value sp) Given two-part message lengths msg1 & msg2, and statistical parameter sp, return a fully parameterised M-Model.
abstract Value	stats(boolean add, Value ss0, Value ss1) Combine sufficient statisticses 'ss0' and 'ss1' additively (add=true), or remove ss1 from ss0 (add=false).
Value	stats(boolean add, Value ss0, Vector ds, int lo, int hi) Combine statisticses ss0 and stats(ds,lo,hi).
Value	stats(Vector ds) Return stats(ds,0,ds.nElts()) (there is more on stats at the link).
abstract Value	stats(Vector ds, int lo, int hi) Return sufficient statistics 'ss' of elements [lo, hi), lo inclusive to hi exclusive, of data-set ds; also see M.stats(...).
java.lang.String	toString() Return a String representation of 'this' UnParameterised Model, including its problem-defining parameters.
UPModel	transform(Function f) transform: (a→b)→UPModel a→UPModel b, convenience function for 'new Transform(f)'.

Methods inherited from class la.la.Function

bOp, fromBop, fromUop, type, uOp

Methods inherited from class la.la.Value

AoM, bOp, compareTo, cons, cts, elt, errMsg, error, force, isTuple, just, n, nElts, nlAoM, pair, print, quad, real, RTE, triple, tuple, UOE, x

Methods inherited from class java.lang.Object

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

Field Detail

dp

protected final Value dp

The problem-defining parameter(s) of 'this' UnParameterised Model; see defnParams().

Constructor Detail

UPModel

public UPModel(Value dp)

Given common-knowledge problem-defining parameter(s), dp, create an UnParameterised Model.

Method Detail

defnParams

public Value defnParams()

Return the common-knowledge, problem-defining parameter(s) for the fully parameterised M-Models that 'this' UPModel can and will produce. This implementation returns dp. Also see Model.statParams().

apply

public Model apply(Value sp)

Given statistical parameters, sp, return a Model, sp2Model(0, 0, sp), i.e., one that is not estimated, having zero part 1 and part 2 message lengths. (Also see Function.apply(la.la.Value).)

Specified by:

apply in class Function.Native

sp2Model

public abstract Model sp2Model(double msg1,
                               double msg2,
                               Value sp)

Given two-part message lengths msg1 & msg2, and statistical parameter sp, return a fully parameterised M-Model. If 'this' UPModel produces a Model m, then sp2Model(m.msg1(), m.msg2(), m.statParams()) must be equivalent to m. Also see apply(sp).

stats

public Value stats(Vector ds)

Return stats(ds,0,ds.nElts()) (there is more on stats at the link). There is unlikely to be any need for a subclass (a UPModel) to override stats(ds).

stats

public abstract Value stats(Vector ds,
                            int lo,
                            int hi)

Return sufficient statistics 'ss' of elements [lo, hi), lo inclusive to hi exclusive, of data-set ds; also see M.stats(...). A simple option is to just return the elements – which is sometimes the best that we can do. Note, this.stats(...) must return statistics that are enough to estimate a fully parameterised M-model, whereas M.stats(...) must be enough to calculate M.nlLH(.); often these are the same, but not always, for example, see BestOf.stats(...). In any case, ss = this.stats(ds,lo,hi) must be compatible with ss2Model(ss) (as must be ss' = M.stats(ds,lo,hi) with M.nlLH(ss')). Weighted data must also be considered. When defining a new UPModel stats(ds,lo,hi) and stats(add,ss0,ss1) are generally what must defined with regard to statistics. The statistics are not necessarily those used under maximum likelihood statistics, say, for example, nlAoM is included for continous Models.

stats

public abstract Value stats(boolean add,
                            Value ss0,
                            Value ss1)

Combine sufficient statisticses 'ss0' and 'ss1' additively (add=true), or remove ss1 from ss0 (add=false). Also see stats(ds,lo,hi).

stats

public Value stats(boolean add,
                   Value ss0,
                   Vector ds,
                   int lo,
                   int hi)

Combine statisticses ss0 and stats(ds,lo,hi). This allows advantage to be taken of additive (incremental) statistics, for example, by Intervals. There is unlikely to be a need for a subclass (a UPModel) to override stats(add,ss0,ds,lo,hi).

estimator

public Estimator estimator(Value ps)

Given parameter(s), 'ps' (possibly parameters of a prior), return an Estimator for 'this' UPModel. Estimator(ps) is optionally implemented and this default throws an Exception. It is desirable that estimator(ps) is implemented (overridden) but we may not have gotten around to it yet, or we may not be clever enough.

toString

public java.lang.String toString()

Return a String representation of 'this' UnParameterised Model, including its problem-defining parameters.

Overrides:

toString in class Function

transform

public UPModel transform(Function f)

transform: (a→b)→UPModel a→UPModel b, convenience function for 'new Transform(f)'. Note that Function 'f' must be 1–to–1. Also see the related but different Model.transform(la.la.Function); note that, as distributions, upm.transform(f)(sp)=upm(sp).transform(f).
The result of this transform(f) is "just" a Model; if you need a more specific sub-class of Model, do as in Continuous.transform(la.la.Function.Cts2Cts).

main

public static void main(java.lang.String[] argv)

Nothing new here; see Test.