analysis/TMVATester.C File Reference

#include "TMVA/Reader.h"
#include "TFile.h"
#include "TTree.h"
#include "TNtuple.h"
#include "TString.h"
#include "TRegexp.h"
#include "TEventList.h"
#include "TLeaf.h"
#include "TCanvas.h"
#include "TH1F.h"
#include "TGraph.h"
#include "TSystem.h"
#include <iostream>
#include <map>
#include <vector>
#include <algorithm>
#include <utility>

Go to the source code of this file.

Macros
#define	MAX   1000
#define	TMVAgettype

Typedefs
typedef std::map< int, int >	CountMap
typedef std::vector< pair < float, int > >	ValueMap

Functions
int	gettype (TTree *t, TString varname)
int	SplitString (TString s, TString delim, TString *toks, int maxtoks)
TString	getFromCut (TString vars)
int	init (TTree *t, TString vars)
TString	getFromWeightFile (TString wfile)
void	TMVATester (TString fname="", TString treename="", TString sigcut="", TString wfile="", TString precut="")

Variables
Float_t	fbranch [MAX]
Float_t	freader [MAX]
Int_t	ibranch [MAX]
Int_t	types [MAX]
Bool_t	bbranch [MAX]
Int_t	ev
Int_t	run
Int_t	mode
Int_t	rec
Int_t	nbranch

Macro Definition Documentation

#define MAX   1000

Definition at line 20 of file analysis/TMVATester.C.

#define TMVAgettype

Definition at line 33 of file analysis/TMVATester.C.

Typedef Documentation

typedef std::map<int, int> CountMap

Definition at line 28 of file analysis/TMVATester.C.

typedef std::vector<pair<float, int> > ValueMap

Definition at line 29 of file analysis/TMVATester.C.

Function Documentation

TString getFromCut

(

TString

vars

)

Definition at line 63 of file analysis/TMVATester.C.

References i, n, res, SplitString(), TString, and v.

{
        TString toks[50];
        int n=SplitString(vars, "&&", toks, 50);
        TRegexp rvar("[_a-zA-Z][_a-zA-Z0-9]*");
        cout <<"n="<<n<<endl;
        TString res="";
        
        for (int i=0;i<n;++i)
        {
                TString v = toks[i](rvar);
                if (v!="")
                {
                        if (v=="tag" || res.Contains(" "+v+" ")) continue;
                        res+=v+" ";
                }
        }
        
        res = res.Strip(TString::kBoth);
        
        return res;
}

TString getFromWeightFile

(

TString

wfile

)

Definition at line 111 of file analysis/TMVATester.C.

References exp(), i, n, r, res, SplitString(), TString, and v.

Referenced by TMVATester().

{
        TString tmp=gSystem->GetFromPipe("grep Expression "+wfile);
        TString v[30];
        int n=SplitString(tmp,"/>",v,30);
        TString res="";
        TRegexp r("Expression=\"[_a-zA-Z0-9]+\"");
        
        for (int i=0;i<n;++i)
        {
                TString exp = (v[i])(r);
                exp.ReplaceAll("Expression=","");
                exp.ReplaceAll("\"","");
                res+=exp+" ";
        }
                
        res = res.Strip(TString::kBoth);
        return res;
}

int gettype	(	TTree *	t,
		TString	varname
	)

Definition at line 34 of file analysis/TMVATester.C.

References TString.

{
        if (t->GetBranch(varname)==0) return -1;
        
        TString leaftype = t->GetLeaf(varname)->GetTypeName();
        
        if (leaftype=="Float_t")         return 0; 
        else if (leaftype=="Int_t")      return 1;  
        else if (leaftype=="Bool_t") return 2;
        
        return -1;
}

int init	(	TTree *	t,
		TString	vars
	)

Definition at line 89 of file analysis/TMVATester.C.

References bbranch, fbranch, gettype(), i, ibranch, SplitString(), TString, types, and v.

{
        TString toks[30];
        int N = SplitString(vars," ",toks,30);
        
        for (int i=0;i<N;++i)
        {
                TString v = toks[i];
                types[i] = gettype(t, v);
                switch (types[i])
                {
                        case 0: t->SetBranchAddress(v, &(fbranch[i])); break;
                        case 1: t->SetBranchAddress(v, &(ibranch[i])); break;
                        case 2: t->SetBranchAddress(v, &(bbranch[i])); break;
                }
                //cout <<toks[i]<<" of type "<<types[i]<<endl;
        }   
        
        return N;
}

int SplitString	(	TString	s,
		TString	delim,
		TString *	toks,
		int	maxtoks
	)

Definition at line 48 of file analysis/TMVATester.C.

References i.

{
        TObjArray *tok = s.Tokenize(delim);
        int N = tok->GetEntries();      
        for (int i=0;i<N;++i) 
                if (i<maxtoks) 
                {
                        toks[i] = ((TObjString*)tok->At(i))->String();
                        toks[i].ReplaceAll("\t","");
                        toks[i] = toks[i].Strip(TString::kBoth);
                }
        return N;
}

void TMVATester	(	TString	fname = "",
		TString	treename = "",
		TString	sigcut = "",
		TString	wfile = "",
		TString	precut = ""
	)

Definition at line 133 of file analysis/TMVATester.C.

References bbranch, c1, elb, els, f, fbranch, freader, getFromCut(), getFromWeightFile(), i, ibranch, init, leg, max(), Nb, Ns, SplitString(), t, TString, types, and vars.

{
        TString vars="";
        
        if ( fname=="" || treename=="" || wfile=="" || sigcut==""  ) 
        {
                cout << "USAGE:\n";
                cout << "TMVATester.C( <input>, <tree>, <sigcut>, <weights>, [precut] )\n\n";
                cout << "   <input>   : input file name containing TTree <tree>\n";
                cout << "   <tree>    : name of the TTree containing signal and background\n";
                cout << "   <sigcut>  : cut separating signal from background -> bgcut = !(sigcut)\n";
                cout << "   <weights> : file containing the weights from training, usually stored in weights/...\n";
                cout << "   [precut]  : optional precut before training; should be the same as for training!'\n\n";
                cout << "EXAMPLE:\n";
                cout << "root -l -b -q 'TMVATester.C(\"demodata.root\",\"ntp\",\"signal>0\",\"weights/demodata_ntp_MLP.weights.xml\",\"\")'\n\n";
                return;
        }
        
        if (vars.Contains("&&")) vars = getFromCut(vars);
        if (vars=="") vars = getFromWeightFile(wfile);
        cout <<"Vars : "<<vars<<endl;
        
        TString bkgcut="!("+sigcut+")";

        if (precut!="")
        {
                cout <<"Precut : "<<precut<<endl;
                sigcut += "&&" + precut;
                bkgcut += "&&" + precut;
        }
        
        TFile *f = TFile::Open(fname);
        TTree *t =(TTree*) f->Get(treename);
        
        // prepare variables and branch mapping
        int Nbr = init(t,vars);
        TString varname[30];
        SplitString(vars," ",varname,30);
        
        // prepare TMVA reader
        TMVA::Reader *reader = new TMVA::Reader("Silent");
        for (int i=0;i<Nbr;++i) reader->AddVariable(varname[i], &freader[i]);
        
        TString mvatype="";
        
        if (wfile.Contains("BDT")) mvatype = "BDT";
        else if (wfile.Contains("MLP")) mvatype = "MLP";
        else if (wfile.Contains("Likelihood")) mvatype = "Likelihood";
        else {cout <<"Unconfigured method in w-file "<<wfile<<endl; return;}
        
        reader->BookMVA(mvatype, wfile);
        
        
        // apply precut cut and create signal and background event lists
        t->Draw(">>els",sigcut);
        t->Draw(">>elb",bkgcut);
        TEventList *els=(TEventList*)gDirectory->Get("els");
        TEventList *elb=(TEventList*)gDirectory->Get("elb");
        
        int Ns = els->GetN();
        int Nb = elb->GetN();
        
        TH1F *hs=new TH1F("hs",Form("%s : output",mvatype.Data()),50,mvatype=="BDT"?-1:0,1);
        TH1F *hb=new TH1F("hb","output background",50,mvatype=="BDT"?-1:0,1);
        hb->SetLineColor(kRed+1);
        hs->SetLineColor(kGreen+2);
        hs->SetLineWidth(2);
        hb->SetLineWidth(2);

        TString friendname = fname;
        friendname.ReplaceAll(".root","_tmvaout.root");
        TFile *ftm = new TFile(friendname,"recreate");
        TNtuple *tf = new TNtuple("tf","tf","tmvaout");

        
        for (int i=0;i<t->GetEntriesFast();++i)
        {
                t->GetEntry(i);
                for (int j=0;j<Nbr;++j) 
                {
                        Float_t var=0;
                        switch (types[j]) {case 2: var = (Float_t)bbranch[j]; break; case 1: var = (Float_t) ibranch[j]; break; default: var=fbranch[j]; }
                        freader[j] = var;
                }
                float resp=reader->EvaluateMVA(mvatype);
                tf->Fill(resp);
        }       
        
        tf->Write();
        ftm->Close();

        // loop through signal event list
        for (int i=0;i<Ns;++i)
        {
                t->GetEntry(els->GetEntry(i));
                
                for (int j=0;j<Nbr;++j) 
                {
                        Float_t var=0;
                        switch (types[j]) {case 2: var = (Float_t)bbranch[j]; break; case 1: var = (Float_t) ibranch[j]; break; default: var=fbranch[j]; }
                        freader[j] = var;
                }
                double resp=reader->EvaluateMVA(mvatype);
                hs->Fill(resp);
        }       
        // loop through backgound event list
        for (int i=0;i<Nb;++i)
        {
                t->GetEntry(elb->GetEntry(i));
                
                for (int j=0;j<Nbr;++j) 
                {
                        Float_t var=0;
                        switch (types[j]) {case 2: var = (Float_t)bbranch[j]; break; case 1: var = (Float_t) ibranch[j]; break; default: var=fbranch[j]; }
                        freader[j] = var;
                }
                double resp=reader->EvaluateMVA(mvatype);
                hb->Fill(resp);
        }       
        
        
        // create the plots
        
        TCanvas *c1=new TCanvas("c1","c1",10,10,1200,600);
        c1->Divide(2,1);
        c1->cd(1);
        hs->Scale(1./hs->GetEntries());
        hb->Scale(1./hb->GetEntries());
        double maxi = max(hs->GetMaximum(), hb->GetMaximum())*1.05;
        
        hs->SetMaximum(maxi);
        hb->SetMaximum(maxi);
        hs->DrawCopy();
        hb->DrawCopy("same");
        
        TLegend *leg=new TLegend(0.3,0.7,0.6,0.85);
        leg->SetBorderSize(0);
        leg->AddEntry(hs, "signal","l");
        leg->AddEntry(hb, "background","l");
        leg->Draw();
        
        c1->cd(2);
        double sums=1., sumb=0.;
        
        // create the ROC curve graph
        TGraph *gr=new TGraph();
        // add points to get a closed curve for Integral
        gr->SetPoint(0,0,0);  
        gr->SetPoint(1,0,1);
        
        for (int i=0;i<=hs->GetNbinsX();++i)
        {
                sums-=hs->GetBinContent(i);
                sumb+=hb->GetBinContent(i);
                gr->SetPoint(i+2,sumb,sums);
        }
        // add point to get a closed curve for Integral
        gr->SetPoint(hs->GetNbinsX()+3,1,0);
        
        // configure graph style
        gr->Draw("ALP");        
        gr->SetMarkerStyle(20);
        gr->SetMarkerSize(0.7);
        gr->GetHistogram()->SetTitle(Form("%s : ROC curve;Background suppression;Signal Efficiency",mvatype.Data()));
        gr->GetHistogram()->SetMinimum(0);
        gr->GetHistogram()->SetMaximum(1.05);
        
        // print the method and integral on the canvas
        TLatex lat;
        lat.SetTextSize(0.045);
        lat.SetNDC();
        lat.SetTextColor(4);
        lat.DrawLatex(0.2,0.2,Form("ROC integral = %.3f",gr->Integral()));
        
        cout <<"Integral of ROC curve (0.5 = useless ... 1.0 = perfect separation): "<<gr->Integral()<<endl;
}

Variable Documentation

Bool_t bbranch[MAX]

Definition at line 24 of file analysis/TMVATester.C.

Int_t ev

Definition at line 26 of file analysis/TMVATester.C.

Float_t fbranch[MAX]

Definition at line 22 of file analysis/TMVATester.C.

Float_t freader[MAX]

Definition at line 22 of file analysis/TMVATester.C.

Int_t ibranch[MAX]

Definition at line 23 of file analysis/TMVATester.C.

Int_t mode

Definition at line 26 of file analysis/TMVATester.C.

Int_t nbranch

Definition at line 26 of file analysis/TMVATester.C.

Int_t rec

Definition at line 26 of file analysis/TMVATester.C.

Int_t run

Definition at line 26 of file analysis/TMVATester.C.

Int_t types[MAX]

Definition at line 23 of file analysis/TMVATester.C.