ROOT logo
ROOT » ROOFIT » ROOSTATS » RooStats::RatioOfProfiledLikelihoodsTestStat

class RooStats::RatioOfProfiledLikelihoodsTestStat: public RooStats::TestStatistic


TestStatistic that returns the ratio of profiled likelihoods.



Function Members (Methods)

Data Members

private:
RooArgSet*fAltPOI
RooStats::ProfileLikelihoodTestStatfAltProfile
RooArgSet*fDetailedOutput
boolfDetailedOutputEnabled
RooStats::ProfileLikelihoodTestStatfNullProfile
Bool_tfSubtractMLE
static Bool_tfgAlwaysReuseNll

Class Charts

Inheritance Inherited Members Includes Libraries
Class Charts

Function documentation

RatioOfProfiledLikelihoodsTestStat()
 Proof constructor. Don't use.
RatioOfProfiledLikelihoodsTestStat(RooAbsPdf& nullPdf, RooAbsPdf& altPdf, const RooArgSet* altPOI = 0)
         Calculates the ratio of profiled likelihoods.

	 By default the calculation is:

	    Lambda(mu_alt , conditional MLE for alt nuisance)
	log --------------------------------------------
   	    Lambda(mu_null , conditional MLE for null nuisance)

	where Lambda is the profile likeihood ratio, so the
	MLE for the null and alternate are subtracted off.

	If SetSubtractMLE(false) then it calculates:

	    L(mu_alt , conditional MLE for alt nuisance)
	log --------------------------------------------
	    L(mu_null , conditional MLE for null nuisance)


	The values of the parameters of interest for the alternative
	hypothesis are taken at the time of the construction.
	If empty, it treats all free parameters as nuisance parameters.

	The value of the parameters of interest for the null hypotheses
	are given at each call of Evaluate(data,nullPOI).

~RatioOfProfiledLikelihoodsTestStat(void)
Double_t ProfiledLikelihood(RooAbsData& data, RooArgSet& poi, RooAbsPdf& pdf)
 returns -logL(poi, conditional MLE of nuisance params)
 it does not subtract off the global MLE
 because  nuisance parameters of null and alternate may not
 be the same.
Double_t Evaluate(RooAbsData& data, RooArgSet& nullParamsOfInterest)
 evaluate the ratio of profile likelihood
void EnableDetailedOutput(bool e = true)
void SetAlwaysReuseNLL(Bool_t flag)
void SetReuseNLL(Bool_t flag)
void SetMinimizer(const char* minimizer)
void SetStrategy(Int_t strategy)
void SetTolerance(Double_t tol)
void SetPrintLevel(Int_t printLevel)
void SetConditionalObservables(const RooArgSet& set)
 set the conditional observables which will be used when creating the NLL
 so the pdf's will not be normalized on the conditional observables when computing the NLL
const RooArgSet* GetDetailedOutput(void)
 Returns detailed output. The value returned by this function is updated after each call to Evaluate().
 The returned RooArgSet contains the following for the alternative and null hypotheses:
 <ul>
 <li> the minimum nll, fitstatus and convergence quality for each fit </li>
 <li> for each fit and for each non-constant parameter, the value, error and pull of the parameter are stored </li>
 </ul>
const TString GetVarName() const
void SetSubtractMLE(bool subtract)
    const bool PValueIsRightTail(void) { return false; } // overwrites default
{fSubtractMLE = subtract;}