Christian,
by default, GetMean, GetRMS return an UNBINNED mean and RMS. You have the full
precision of the original points.
Rene Brun
Christian Holm Christensen wrote:
>
> Hi Rene,
>
> On Wed, 20 Mar 2002 12:18:50 +0000
> Rene Brun <Rene.Brun@cern.ch> wrote
> concerning "Re: [ROOT] Getting mean of Ntuple variables":
> > Hi Christian,
> >
> > A TTree::GetMean,RMS could be implemented. I am however reluctant to
> > implement these functions. Very likely you would like a
> >
> > GetMean(variable, selection)
> >
> > and also the full machinery with
> >
> > GetMean(const char *varexp, const char *selection,
> > Int_t nentries=1000000000, Int_t firstentry=0);
> >
> > etc for GetRMS.
> >
> > This functionality is already provided by TTree::Draw as indicated
> > by Pasha. I believe it is more general to do
> >
> > tree.Draw("var>>hist,"some selection",etc);
> > hist->getMean();
>
> It does require that you make a histogram, which is not necessarily
> what you want, especially not in batch processing, and the like. And
> sometimes I guess you really would like an unbinned mean and RMS. I
> really see no reason why not to give the users the possiblity of
> getting the mean and RMS directly for the horses mouth (sample), also
> in that it makes it easier to make various kinds of statistical
> analysis on the trees.
>
> Yours,
>
> Christian Holm Christensen -------------------------------------------
> Address: Sankt Hansgade 23, 1. th. Phone: (+45) 35 35 96 91
> DK-2200 Copenhagen N Cell: (+45) 28 82 16 23
> Denmark Office: (+45) 353 25 305
> Email: cholm@nbi.dk Web: www.nbi.dk/~cholm
>
> >
> > Rene Brun
> >
> >
> > Christian Holm Christensen wrote:
> > >
> > > Hi,
> > >
> > > On Wed, 13 Mar 2002 13:19:55 -0800
> > > "Aron, Navneet" <navneet@SLAC.stanford.edu> wrote
> > > concerning "[ROOT] Getting mean of Ntuple variables":
> > > >
> > > > Hi,
> > > > Can any one tell me how do we get the mean , rms values of the various ntuple variables in the root file.
> > > > My try is as follows:
> > > >
> > > > {
> > > > gROOT->Reset();
> > > > TFile *r = new TFile("g1onaxis.root");
> > > > TTree *tree = (TTree*)r->Get("TKR_Hits_In_Lyr_0");
> > > > Float_t mean = tree->GetMean();
> > > > }
> > >
> > > While the suggestion given by Pasha is indeed the right way to go, it
> > > would be nice to have methods like
> > >
> > > Double_t TTree::GetMean(const Char_t* leaf)
> > > Double_t TTree::GetRMS(const Char_t* leaf)
> > >
> > > (should not be const, since the tree should cache the calculation),
> > > and maybe higher moments too.
> > >
> > > I once made a small class TStatistics, which can help you do low-level
> > > statistics (average, variance, covariance) on a N-dimensional sample.
> > > The advantage of this class is, that it minimizes the rounding errors
> > > due to finite machine precision, _and_ you only need to loop over the
> > > data once to make the (co)variance. The algorithm is the same as the
> > > one used in TPrincipal. It also has a service (static) method for a
> > > 1D sample, so that one can easily use it to caculate the average and
> > > variance of a leaf variable:
> > >
> > > TTree* tree = (TTree*)gDirectory->Get("Tree");
> > > Double_t x;
> > > Double_t mean = 0;
> > > Double_t rms = 0;
> > > tree->SetBranchAddress("X", &x);
> > > Ssiz_t n = tree->GetEntries();
> > > for (Int_t i = 0; i < Int_t(n); i++) {
> > > tree->GetEntry(i);
> > > TStatistics::AddPoint(x, i, mean, rms);
> > > }
> > >
> > > Notice, that the average and variance is calculated unbinned, unlike
> > > in a histogram.
> > >
> > > You can get the TStatistics class from my web site [1].
> > >
> > > The tar-ball is standalone, but also demonstrates how to use Autotools
> > > with ROOT.
> > >
> > > Yours,
> > >
> > > Christian Holm Christensen -------------------------------------------
> > > Address: Sankt Hansgade 23, 1. th. Phone: (+45) 35 35 96 91
> > > DK-2200 Copenhagen N Cell: (+45) 28 82 16 23
> > > Denmark Office: (+45) 353 25 305
> > > Email: cholm@nbi.dk Web: www.nbi.dk/~cholm
> > >
> > >
> > > [1] http://cholm.home.cern.ch/cholm/root/#tstatistics
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:50:46 MET