rvl.stat.anova
Class Model

java.lang.Object
  rvl.stat.anova.Model

public class Model
extends java.lang.Object

Field Summary

boolean

recalcLU

Constructor Summary

Model()

Model(java.lang.String s)
Construct a Model by parsing a string s

Method Summary

void	addFactor(Factor f)
void	addFactor(Factor f, boolean expand)
void	addTerm(Term t) Add a term to the model
double[][]	EMSCoefs()
java.lang.String	EMSString()
java.util.Vector	getAllCompRestr(Term compTerm)
double[]	getCompCoefs(Term compTerm, Factor[] restr)
double[]	getCompErrorTerms(double[] coef)
java.util.Vector	getCompRestr(Term compTerm)
double[]	getCompVariance(Term compTerm, Factor[] restr, double[] effSD)
java.lang.String[]	getCompVarString(Term compTerm, Factor[] restr)
double[][]	getErrorTerms()
double[][]	getErrorTerms(double[][] C)
Factor	getFac(int i)
Factor	getFac(java.lang.String s)
int	getNobs()
double[]	getPowerInfo(int i)
Term	getTerm(int i)
int	nFac()
int	nTerm()
double[]	power(double[] sd, double alpha)
void	printEMS()
void	removeTerm(int i)
void	setFixed(java.lang.String s) Set specified factors to be fixed
void	setLevels(java.lang.String s) Set numbers of levels based on a string with format name #levels name #levels ...
void	setRandom(java.lang.String s) Set specified factors to be random
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

recalcLU

public boolean recalcLU

Constructor Detail

Model

public Model()

Model

public Model(java.lang.String s)

Construct a Model by parsing a string s

Method Detail

getFac

public Factor getFac(int i)

Returns:

ith factor in model

getFac

public Factor getFac(java.lang.String s)

Returns:

factor named s in the model (case is ignored) Note: Nesting factors are ignored, and s is assumed not to have any parenthesized expressions.

getTerm

public Term getTerm(int i)

nFac

public int nFac()

nTerm

public int nTerm()

addFactor

public void addFactor(Factor f)

addFactor

public void addFactor(Factor f,
                      boolean expand)

addTerm

public void addTerm(Term t)

Add a term to the model

removeTerm

public void removeTerm(int i)

setLevels

public void setLevels(java.lang.String s)

Set numbers of levels based on a string with format name #levels name #levels ... (space-delimited) Extensions: 1. Factors can be locked to have common numbers of levels by specifying name1=name2=... instead of a single name. 2. A factor's levels may be used to define a fractional experiment by preceding its name with a "/" Example (3-period crossover design): Pass the string /seq=per=drug 3 SUBJ 5

setRandom

public void setRandom(java.lang.String s)

Set specified factors to be random

setFixed

public void setFixed(java.lang.String s)

Set specified factors to be fixed

getNobs

public int getNobs()

Returns:

the total number of observations in the model

EMSCoefs

public double[][] EMSCoefs()

Returns:

matrix of coefficients of the EMSs The ith row contains the coefficients of the variance components. This linear combination is the EMS of the ith term

getErrorTerms

public double[][] getErrorTerms(double[][] C)

Returns:

matrix of coefficients of the mean squares comprising the error terms for each term

getErrorTerms

public double[][] getErrorTerms()

power

public double[] power(double[] sd,
                      double alpha)

Returns:

powers of tests, given the SD components in sd[] and test size alpha
Note: If test is invalid, a negative power is returned: A "power" of -1 means no error term; -2 means denominator df are too low; -3 means an arithmetic exception occurred.

getPowerInfo

public double[] getPowerInfo(int i)

Returns:

basic information for use in power computation for the i-th term: { leading coef, EMS(denom), numdf, dendf }
These values are stored in the last call to power(), so you should call power() first if it hasn't benn called previously, or if anything has changed.

printEMS

public void printEMS()

EMSString

public java.lang.String EMSString()

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getCompRestr

public java.util.Vector getCompRestr(Term compTerm)

Returns:

Vector of fac[] arrays corresponding to restrictions on comparisons of levels or factor combinations in compTerm. Each restriction represents a combination of factors that, if held fixed in a comparison, will have a smaller variance than just any comparison. Note: The first element is always null, for an unrestricted comparison.
compTerm should be fixed -- otherwise null is returned

getAllCompRestr

public java.util.Vector getAllCompRestr(Term compTerm)

Returns:

all possible restrictions that could be placed on a comparison. This is just like getCompRestr(), except it does not confine itself to the restrictions that make a difference in the variance of a comparison or contrast.

getCompVariance

public double[] getCompVariance(Term compTerm,
                                Factor[] restr,
                                double[] effSD)

Returns:

["base variance", d.f.] of a comparison of the levels of compTerm, when it is restricted to the same levels of factors in restr, given the SDs in effSD (note - only the components of effSD that correspond to random terms are used.
Multiply this result by the sum of squares of the contrast coefficients to get the actual variance.

getCompCoefs

public double[] getCompCoefs(Term compTerm,
                             Factor[] restr)

Returns:

the coefficients of the variance components to obtain the "base variance" described above

getCompErrorTerms

public double[] getCompErrorTerms(double[] coef)

Returns:

lin. comb. of MSs that estimates the given lin. comb. of variances in coef[].
If model has changed since last call, set recalcLU to true before calling

getCompVarString

public java.lang.String[] getCompVarString(Term compTerm,
                                           Factor[] restr)

Returns:

strings that describes the variance of a comparison.
Element 0 is the combination of variances (to be multiplied by sum of squares of contrast coefs);
Element 1 is the combination of mean squares to estimate this variance.

Should set recalcLU to true if model has changed since last call.