Logo ROOT  
Reference Guide
 
Loading...
Searching...
No Matches
TransformationHandler.cxx
Go to the documentation of this file.
1// @(#)root/tmva $Id$
2// Author: Andreas Hoecker, Peter Speckmayer, Joerg Stelzer, Helge Voss, Eckhard von Toerne, Jan Therhaag
3
4/**********************************************************************************
5 * Project: TMVA - a Root-integrated toolkit for multivariate data analysis *
6 * Package: TMVA *
7 * Class : TransformationHandler *
8 * *
9 * *
10 * Description: *
11 * Implementation (see header for description) *
12 * *
13 * Authors (alphabetical): *
14 * Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
15 * Peter Speckmayer <speckmay@mail.cern.ch> - CERN, Switzerland *
16 * Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
17 * Jan Therhaag <Jan.Therhaag@cern.ch> - U of Bonn, Germany *
18 * Eckhard v. Toerne <evt@uni-bonn.de> - U of Bonn, Germany *
19 * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
20 * *
21 * Copyright (c) 2005-2011: *
22 * CERN, Switzerland *
23 * MPI-K Heidelberg, Germany *
24 * U. of Bonn, Germany *
25 * *
26 * Redistribution and use in source and binary forms, with or without *
27 * modification, are permitted according to the terms listed in LICENSE *
28 * (see tmva/doc/LICENSE) *
29 **********************************************************************************/
30
31/*! \class TMVA::TransformationHandler
32\ingroup TMVA
33Class that contains all the data information.
34*/
35
37
38#include "TMVA/Config.h"
39#include "TMVA/DataSet.h"
40#include "TMVA/DataSetInfo.h"
41#include "TMVA/Event.h"
42#include "TMVA/MsgLogger.h"
43#include "TMVA/Ranking.h"
44#include "TMVA/Tools.h"
45#include "TMVA/Types.h"
48#include "TMVA/VariableInfo.h"
54
55#include "TAxis.h"
56#include "TDirectory.h"
57#include "TH1.h"
58#include "TH2.h"
59#include "TList.h"
60#include "TMath.h"
61#include "TProfile.h"
62
63#include <vector>
64#include <iomanip>
65
66////////////////////////////////////////////////////////////////////////////////
67/// constructor
68
70 : fDataSetInfo(dsi),
71 fRootBaseDir(0),
72 fCallerName (callerName),
73 fLogger ( new MsgLogger(TString("TFHandler_" + callerName).Data(), kINFO) )
74{
75 // produce one entry for each class and one entry for all classes. If there is only one class,
76 // produce only one entry
77 fNumC = (dsi.GetNClasses()<= 1) ? 1 : dsi.GetNClasses()+1;
78
79 fVariableStats.resize( fNumC );
80 for (Int_t i=0; i<fNumC; i++ ) fVariableStats.at(i).resize(dsi.GetNVariables() + dsi.GetNTargets());
81}
82
83////////////////////////////////////////////////////////////////////////////////
84/// destructor
85
87{
88 std::vector<Ranking*>::const_iterator it = fRanking.begin();
89 for (; it != fRanking.end(); ++it) delete *it;
90
91 fTransformations.SetOwner();
92 delete fLogger;
93}
94
95////////////////////////////////////////////////////////////////////////////////
96
98{
99 fCallerName = name;
100 fLogger->SetSource( TString("TFHandler_" + fCallerName).Data() );
101}
102
103////////////////////////////////////////////////////////////////////////////////
104
106{
107 TString tfname = trf->Log().GetName();
108 trf->Log().SetSource(TString(fCallerName+"_"+tfname+"_TF").Data());
109 fTransformations.Add(trf);
110 fTransformationsReferenceClasses.push_back( cls );
111 return trf;
112}
113
114////////////////////////////////////////////////////////////////////////////////
115/// Caches calculated summary statistics of transformed variables.
116///
117/// \param[in] k index of class
118/// \param[in] ivar index of variable
119/// \param[in] mean the mean value of the variable
120/// \param[in] rms the root-mean-square value of the variable
121/// \param[in] min the minimum value of the variable
122/// \param[in] max the maximum value of the variable
123
125{
126 if (rms <= 0 || TMath::IsNaN(rms)) {
127 Log() << kWARNING << "Variable \"" << Variable(ivar).GetExpression()
128 << "\" has zero, negative, or NaN RMS^2: "
129 << rms
130 << " ==> set to zero. Please check the variable content" << Endl;
131 rms = 0;
132 }
133
134 VariableStat stat; stat.fMean = mean; stat.fRMS = rms; stat.fMin = min; stat.fMax = max;
135 fVariableStats.at(k).at(ivar) = stat;
136}
137
138////////////////////////////////////////////////////////////////////////////////
139/// overrides the setting for all classes! (this is put in basically for the likelihood-method)
140/// be careful with the usage this method
141
143{
144 for (UInt_t i = 0; i < fTransformationsReferenceClasses.size(); i++) {
145 fTransformationsReferenceClasses.at( i ) = cls;
146 }
147}
148
149////////////////////////////////////////////////////////////////////////////////
150/// the transformation
151
153{
154 TListIter trIt(&fTransformations);
155 std::vector<Int_t>::const_iterator rClsIt = fTransformationsReferenceClasses.begin();
156 const Event* trEv = ev;
157 while (VariableTransformBase *trf = (VariableTransformBase*) trIt()) {
158 if (rClsIt == fTransformationsReferenceClasses.end()) Log() << kFATAL<< "invalid read in TransformationHandler::Transform " <<Endl;
159 trEv = trf->Transform(trEv, (*rClsIt) );
160 ++rClsIt;
161 }
162 return trEv;
163}
164
165////////////////////////////////////////////////////////////////////////////////
166
167const TMVA::Event* TMVA::TransformationHandler::InverseTransform( const Event* ev, Bool_t suppressIfNoTargets ) const
168{
169 if (fTransformationsReferenceClasses.empty()){
170 //Log() << kWARNING << __FILE__ <<":InverseTransform fTransformationsReferenceClasses is empty" << Endl;
171 return ev;
172 }
173 // the inverse transformation
174 TListIter trIt(&fTransformations, kIterBackward);
175 std::vector< Int_t >::const_iterator rClsIt = fTransformationsReferenceClasses.end();
176 std::vector<Int_t>::const_iterator rClsBegin = fTransformationsReferenceClasses.begin();
177 --rClsIt;
178 const Event* trEv = ev;
179 UInt_t nvars = 0, ntgts = 0, nspcts = 0;
180 while (VariableTransformBase *trf = (VariableTransformBase*) trIt() ) { // shouldn't be the transformation called in the inverse order for the inversetransformation?????
181 if (trf->IsCreated()) {
182 trf->CountVariableTypes( nvars, ntgts, nspcts );
183 if( !(suppressIfNoTargets && ntgts==0) )
184 trEv = trf->InverseTransform(ev, (*rClsIt) );
185 }
186 else break;
187 if (rClsIt > rClsBegin)
188 --rClsIt;
189 }
190 return trEv;
191
192
193 // TListIter trIt(&fTransformations);
194 // std::vector< Int_t >::const_iterator rClsIt = fTransformationsReferenceClasses.begin();
195 // const Event* trEv = ev;
196 // UInt_t nvars = 0, ntgts = 0, nspcts = 0;
197 // while (VariableTransformBase *trf = (VariableTransformBase*) trIt() ) { // shouldn't be the transformation called in the inverse order for the inversetransformation?????
198 // if (trf->IsCreated()) {
199 // trf->CountVariableTypes( nvars, ntgts, nspcts );
200 // if( !(suppressIfNoTargets && ntgts==0) )
201 // trEv = trf->InverseTransform(ev, (*rClsIt) );
202 // }
203 // else break;
204 // rClsIt++;
205 // }
206 // return trEv;
207
208}
209
210////////////////////////////////////////////////////////////////////////////////
211/// computation of transformation
212
213const std::vector<TMVA::Event*>* TMVA::TransformationHandler::CalcTransformations( const std::vector<Event*>& events,
214 Bool_t createNewVector )
215{
216 if (fTransformations.GetEntries() <= 0)
217 return &events;
218
219 // the transformedEvents are initialised with the initial events and then
220 // subsequently replaced with transformed ones. The n-th transformation will
221 // and on the events as they look like after the (n-1)-the transformation
222 // as intended for the chained transformations
223 std::vector<Event*> *transformedEvents = new std::vector<TMVA::Event*>(events.size());
224 for ( UInt_t ievt = 0; ievt<events.size(); ievt++)
225 transformedEvents->at(ievt) = new Event(*events.at(ievt));
226
227 TListIter trIt(&fTransformations);
228 std::vector< Int_t >::iterator rClsIt = fTransformationsReferenceClasses.begin();
229 while (VariableTransformBase *trf = (VariableTransformBase*) trIt()) {
230 if (trf->PrepareTransformation(*transformedEvents)) {
231 for (UInt_t ievt = 0; ievt<transformedEvents->size(); ievt++) { // loop through all events
232 *(*transformedEvents)[ievt] = *trf->Transform((*transformedEvents)[ievt],(*rClsIt));
233 }
234 ++rClsIt;
235 }
236 }
237
238 CalcStats(*transformedEvents);
239
240 // plot the variables once in this transformation
241 PlotVariables(*transformedEvents);
242
243 //sometimes, the actual transformed event vector is not used for anything but the previous
244 //CalcStat and PlotVariables calles, in that case, we delete it again (and return NULL)
245 if (!createNewVector) { // if we don't want that newly created event vector to persist, then delete it
246 for ( UInt_t ievt = 0; ievt<transformedEvents->size(); ievt++)
247 delete (*transformedEvents)[ievt];
248 delete transformedEvents;
249 transformedEvents=NULL;
250 }
251
252 return transformedEvents; // give back the newly created event collection (containing the transformed events)
253}
254
255////////////////////////////////////////////////////////////////////////////////
256/// method to calculate minimum, maximum, mean, and RMS for all
257/// variables used in the MVA
258
259void TMVA::TransformationHandler::CalcStats (const std::vector<Event*>& events )
260{
261 UInt_t nevts = events.size();
262
263 if (nevts==0)
264 Log() << kFATAL << "No events available to find min, max, mean and rms" << Endl;
265
266 // if transformation has not been succeeded, the tree may be empty
267 const UInt_t nvar = events[0]->GetNVariables();
268 const UInt_t ntgt = events[0]->GetNTargets();
269
270 Double_t *sumOfWeights = new Double_t[fNumC];
271 Double_t* *x2 = new Double_t*[fNumC];
272 Double_t* *x0 = new Double_t*[fNumC];
273 Double_t* *varMin = new Double_t*[fNumC];
274 Double_t* *varMax = new Double_t*[fNumC];
275
276 for (Int_t cls=0; cls<fNumC; cls++) {
277 sumOfWeights[cls]=0;
278 x2[cls] = new Double_t[nvar+ntgt];
279 x0[cls] = new Double_t[nvar+ntgt];
280 varMin[cls] = new Double_t[nvar+ntgt];
281 varMax[cls] = new Double_t[nvar+ntgt];
282 for (UInt_t ivar=0; ivar<nvar+ntgt; ivar++) {
283 x0[cls][ivar] = x2[cls][ivar] = 0;
284 varMin[cls][ivar] = DBL_MAX;
285 varMax[cls][ivar] = -DBL_MAX;
286 }
287 }
288
289 for (UInt_t ievt=0; ievt<nevts; ievt++) {
290 const Event* ev = events[ievt];
291 Int_t cls = ev->GetClass();
292
293 Double_t weight = ev->GetWeight();
294 sumOfWeights[cls] += weight;
295 if (fNumC > 1 ) sumOfWeights[fNumC-1] += weight; // if more than one class, store values for all classes
296 for (UInt_t var_tgt = 0; var_tgt < 2; var_tgt++ ){ // first for variables, then for targets
297 UInt_t nloop = ( var_tgt==0?nvar:ntgt );
298 for (UInt_t ivar=0; ivar<nloop; ivar++) {
299 Double_t x = ( var_tgt==0?ev->GetValue(ivar):ev->GetTarget(ivar) );
300
301 if (x < varMin[cls][(var_tgt*nvar)+ivar]) varMin[cls][(var_tgt*nvar)+ivar]= x;
302 if (x > varMax[cls][(var_tgt*nvar)+ivar]) varMax[cls][(var_tgt*nvar)+ivar]= x;
303
304 x0[cls][(var_tgt*nvar)+ivar] += x*weight;
305 x2[cls][(var_tgt*nvar)+ivar] += x*x*weight;
306
307 if (fNumC > 1) {
308 if (x < varMin[fNumC-1][(var_tgt*nvar)+ivar]) varMin[fNumC-1][(var_tgt*nvar)+ivar]= x;
309 if (x > varMax[fNumC-1][(var_tgt*nvar)+ivar]) varMax[fNumC-1][(var_tgt*nvar)+ivar]= x;
310
311 x0[fNumC-1][(var_tgt*nvar)+ivar] += x*weight;
312 x2[fNumC-1][(var_tgt*nvar)+ivar] += x*x*weight;
313 }
314 }
315 }
316 }
317
318
319 // set Mean and RMS
320 for (UInt_t var_tgt = 0; var_tgt < 2; var_tgt++ ){ // first for variables, then for targets
321 UInt_t nloop = ( var_tgt==0?nvar:ntgt );
322 for (UInt_t ivar=0; ivar<nloop; ivar++) {
323 for (Int_t cls = 0; cls < fNumC; cls++) {
324 Double_t mean = x0[cls][(var_tgt*nvar)+ivar]/sumOfWeights[cls];
325 Double_t rms = TMath::Sqrt( x2[cls][(var_tgt*nvar)+ivar]/sumOfWeights[cls] - mean*mean);
326 AddStats(cls, (var_tgt*nvar)+ivar, mean, rms, varMin[cls][(var_tgt*nvar)+ivar], varMax[cls][(var_tgt*nvar)+ivar]);
327 }
328 }
329 }
330
331 // ------ pretty output of basic statistics -------------------------------
332 // find maximum length in V (and column title)
333 UInt_t maxL = 8, maxV = 0;
334 std::vector<UInt_t> vLengths;
335 for (UInt_t ivar=0; ivar<nvar+ntgt; ivar++) {
336 if( ivar < nvar )
337 maxL = TMath::Max( (UInt_t)Variable(ivar).GetLabel().Length(), maxL );
338 else
339 maxL = TMath::Max( (UInt_t)Target(ivar-nvar).GetLabel().Length(), maxL );
340 }
341 maxV = maxL + 2;
342 // full column length
343 UInt_t clen = maxL + 4*maxV + 11;
344 Log() << kHEADER ;
345 //for (UInt_t i=0; i<clen; i++) //Log() << "-";
346
347 //Log() << Endl;
348 // full column length
349 Log() << std::setw(maxL) << "Variable";
350 Log() << " " << std::setw(maxV) << "Mean";
351 Log() << " " << std::setw(maxV) << "RMS";
352 Log() << " " << std::setw(maxV) << "[ Min ";
353 Log() << " " << std::setw(maxV) << " Max ]"<< Endl;;
354 for (UInt_t i=0; i<clen; i++) Log() << "-";
355 Log() << Endl;
356
357 // the numbers
358 TString format = "%#11.5g";
359 for (UInt_t ivar=0; ivar<nvar+ntgt; ivar++) {
360 if( ivar < nvar )
361 Log() << std::setw(maxL) << Variable(ivar).GetLabel() << ":";
362 else
363 Log() << std::setw(maxL) << Target(ivar-nvar).GetLabel() << ":";
364 Log() << std::setw(maxV) << Form( format.Data(), GetMean(ivar) );
365 Log() << std::setw(maxV) << Form( format.Data(), GetRMS(ivar) );
366 Log() << " [" << std::setw(maxV) << Form( format.Data(), GetMin(ivar) );
367 Log() << std::setw(maxV) << Form( format.Data(), GetMax(ivar) ) << " ]";
368 Log() << Endl;
369 }
370 for (UInt_t i=0; i<clen; i++) Log() << "-";
371 Log() << Endl;
372 // ------------------------------------------------------------------------
373
374 delete[] sumOfWeights;
375 for (Int_t cls=0; cls<fNumC; cls++) {
376 delete [] x2[cls];
377 delete [] x0[cls];
378 delete [] varMin[cls];
379 delete [] varMax[cls];
380 }
381 delete [] x2;
382 delete [] x0;
383 delete [] varMin;
384 delete [] varMax;
385}
386
387////////////////////////////////////////////////////////////////////////////////
388/// create transformation function
389
390void TMVA::TransformationHandler::MakeFunction( std::ostream& fout, const TString& fncName, Int_t part ) const
391{
392 TListIter trIt(&fTransformations);
393 std::vector< Int_t >::const_iterator rClsIt = fTransformationsReferenceClasses.begin();
394 UInt_t trCounter=1;
395 while (VariableTransformBase *trf = (VariableTransformBase*) trIt() ) {
396 trf->MakeFunction(fout, fncName, part, trCounter++, (*rClsIt) );
397 ++rClsIt;
398 }
399 if (part==1) {
400 for (Int_t i=0; i<fTransformations.GetSize(); i++) {
401 fout << " void InitTransform_"<<i+1<<"();" << std::endl;
402 fout << " void Transform_"<<i+1<<"( std::vector<double> & iv, int sigOrBgd ) const;" << std::endl;
403 }
404 }
405 if (part==2) {
406 fout << std::endl;
407 fout << "//_______________________________________________________________________" << std::endl;
408 fout << "inline void " << fncName << "::InitTransform()" << std::endl;
409 fout << "{" << std::endl;
410 for (Int_t i=0; i<fTransformations.GetSize(); i++)
411 fout << " InitTransform_"<<i+1<<"();" << std::endl;
412 fout << "}" << std::endl;
413 fout << std::endl;
414 fout << "//_______________________________________________________________________" << std::endl;
415 fout << "inline void " << fncName << "::Transform( std::vector<double>& iv, int sigOrBgd ) const" << std::endl;
416 fout << "{" << std::endl;
417 for (Int_t i=0; i<fTransformations.GetSize(); i++)
418 fout << " Transform_"<<i+1<<"( iv, sigOrBgd );" << std::endl;
419
420 fout << "}" << std::endl;
421 }
422}
423
424////////////////////////////////////////////////////////////////////////////////
425/// return transformation name
426
428{
429 TString name("Id");
430 TListIter trIt(&fTransformations);
432 if ((trf = (VariableTransformBase*) trIt())) {
433 name = TString(trf->GetShortName());
434 while ((trf = (VariableTransformBase*) trIt())) name += "_" + TString(trf->GetShortName());
435 }
436 return name;
437}
438
439////////////////////////////////////////////////////////////////////////////////
440/// incorporates transformation type into title axis (usually for histograms)
441
443{
444 TString xtit = info.GetTitle();
445 // indicate transformation, but not in case of single identity transform
446 if (fTransformations.GetSize() >= 1) {
447 if (fTransformations.GetSize() > 1 ||
448 ((VariableTransformBase*)GetTransformationList().Last())->GetVariableTransform() != Types::kIdentity) {
449 xtit += " (" + GetName() + ")";
450 }
451 }
452 return xtit;
453}
454
455
456////////////////////////////////////////////////////////////////////////////////
457/// create histograms from the input variables
458/// - histograms for all input variables
459/// - scatter plots for all pairs of input variables
460
461void TMVA::TransformationHandler::PlotVariables (const std::vector<Event*>& events, TDirectory* theDirectory )
462{
463 if (fRootBaseDir==0 && theDirectory == 0) return;
464
465 Log() << kDEBUG << "Plot event variables for ";
466 if (theDirectory !=0) Log()<< TString(theDirectory->GetName()) << Endl;
467 else Log() << GetName() << Endl;
468
469 // extension for transformation type
470 TString transfType = "";
471 if (theDirectory == 0) {
472 transfType += "_";
473 transfType += GetName();
474 }else{ // you plot for the individual classifiers. Note, here the "statistics" still need to be calculated as you are in the testing phase
475 CalcStats(events);
476 }
477
478 const UInt_t nvar = fDataSetInfo.GetNVariables();
479 const UInt_t ntgt = fDataSetInfo.GetNTargets();
480 const Int_t ncls = fDataSetInfo.GetNClasses();
481
482 // Create all histograms
483 // do both, scatter and profile plots
484 std::vector<std::vector<TH1*> > hVars( ncls ); // histograms for variables
485 std::vector<std::vector<std::vector<TH2F*> > > mycorr( ncls ); // histograms for correlations
486 std::vector<std::vector<std::vector<TProfile*> > > myprof( ncls ); // histograms for profiles
487
488 for (Int_t cls = 0; cls < ncls; cls++) {
489 hVars.at(cls).resize ( nvar+ntgt );
490 hVars.at(cls).assign ( nvar+ntgt, 0 ); // fill with zeros
491 mycorr.at(cls).resize( nvar+ntgt );
492 myprof.at(cls).resize( nvar+ntgt );
493 for (UInt_t ivar=0; ivar < nvar+ntgt; ivar++) {
494 mycorr.at(cls).at(ivar).resize( nvar+ntgt );
495 myprof.at(cls).at(ivar).resize( nvar+ntgt );
496 mycorr.at(cls).at(ivar).assign( nvar+ntgt, 0 ); // fill with zeros
497 myprof.at(cls).at(ivar).assign( nvar+ntgt, 0 ); // fill with zeros
498 }
499 }
500
501 // if there are too many input variables, the creation of correlations plots blows up
502 // memory and basically kills the TMVA execution
503 // --> avoid above critical number (which can be user defined)
504 if (nvar+ntgt > (UInt_t)gConfig().GetVariablePlotting().fMaxNumOfAllowedVariablesForScatterPlots) {
505 Int_t nhists = (nvar+ntgt)*(nvar+ntgt - 1)/2;
506 Log() << kINFO << gTools().Color("dgreen") << Endl;
507 Log() << kINFO << "<PlotVariables> Will not produce scatter plots ==> " << Endl;
508 Log() << kINFO
509 << "| The number of " << nvar << " input variables and " << ntgt << " target values would require "
510 << nhists << " two-dimensional" << Endl;
511 Log() << kINFO
512 << "| histograms, which would occupy the computer's memory. Note that this" << Endl;
513 Log() << kINFO
514 << "| suppression does not have any consequences for your analysis, other" << Endl;
515 Log() << kINFO
516 << "| than not disposing of these scatter plots. You can modify the maximum" << Endl;
517 Log() << kINFO
518 << "| number of input variables allowed to generate scatter plots in your" << Endl;
519 Log() << "| script via the command line:" << Endl;
520 Log() << kINFO
521 << "| \"(TMVA::gConfig().GetVariablePlotting()).fMaxNumOfAllowedVariablesForScatterPlots = <some int>;\""
522 << gTools().Color("reset") << Endl;
523 Log() << Endl;
524 Log() << kINFO << "Some more output" << Endl;
525 }
526
530
531 for (UInt_t var_tgt = 0; var_tgt < 2; var_tgt++) { // create the histos first for the variables, then for the targets
532 UInt_t nloops = ( var_tgt == 0? nvar:ntgt ); // number of variables or number of targets
533 for (UInt_t ivar=0; ivar<nloops; ivar++) {
534 const VariableInfo& info = ( var_tgt == 0 ? Variable( ivar ) : Target(ivar) ); // choose the appropriate one (variable or target)
535 TString myVari = info.GetInternalName();
536
537 Double_t mean = fVariableStats.at(fNumC-1).at( ( var_tgt*nvar )+ivar).fMean;
538 Double_t rms = fVariableStats.at(fNumC-1).at( ( var_tgt*nvar )+ivar).fRMS;
539
540 for (Int_t cls = 0; cls < ncls; cls++) {
541
542 TString className = fDataSetInfo.GetClassInfo(cls)->GetName();
543
544 // add "target" in case of target variable (required for plotting macros)
545 className += (ntgt == 1 && var_tgt == 1 ? "_target" : "");
546
547 // choose reasonable histogram ranges, by removing outliers
548 TH1* h = 0;
549 if (info.GetVarType() == 'I') {
550 // special treatment for integer variables
551 Int_t xmin = TMath::Nint( GetMin( ( var_tgt*nvar )+ivar) );
552 Int_t xmax = TMath::Nint( GetMax( ( var_tgt*nvar )+ivar) + 1 );
553 Int_t nbins = xmax - xmin;
554
555 h = new TH1F( TString::Format("%s__%s%s", myVari.Data(), className.Data(), transfType.Data()).Data(),
556 info.GetTitle(), nbins, xmin, xmax );
557 }
558 else {
559 Double_t xmin = TMath::Max( GetMin( ( var_tgt*nvar )+ivar), mean - timesRMS*rms );
560 Double_t xmax = TMath::Min( GetMax( ( var_tgt*nvar )+ivar), mean + timesRMS*rms );
561
562 //std::cout << "Class="<<cls<<" xmin="<<xmin << " xmax="<<xmax<<" mean="<<mean<<" rms="<<rms<<" timesRMS="<<timesRMS<<std::endl;
563 // protection
564 if (xmin >= xmax) xmax = xmin*1.1; // try first...
565 if (xmin >= xmax) xmax = xmin + 1; // this if xmin == xmax == 0
566 // safety margin for values equal to the maximum within the histogram
567 xmax += (xmax - xmin)/nbins1D;
568
569 h = new TH1F( TString::Format("%s__%s%s", myVari.Data(), className.Data(), transfType.Data()).Data(),
570 info.GetTitle(), nbins1D, xmin, xmax );
571 }
572
573 h->GetXaxis()->SetTitle( gTools().GetXTitleWithUnit( GetVariableAxisTitle( info ), info.GetUnit() ) );
574 h->GetYaxis()->SetTitle( gTools().GetYTitleWithUnit( *h, info.GetUnit(), kFALSE ) );
575 hVars.at(cls).at((var_tgt*nvar)+ivar) = h;
576
577 // profile and scatter plots
578 if (nvar+ntgt <= (UInt_t)gConfig().GetVariablePlotting().fMaxNumOfAllowedVariablesForScatterPlots) {
579
580 for (UInt_t v_t = 0; v_t < 2; v_t++) {
581 UInt_t nl = ( v_t==0?nvar:ntgt );
582 UInt_t start = ( v_t==0? (var_tgt==0?ivar+1:0):(var_tgt==0?nl:ivar+1) );
583 for (UInt_t j=start; j<nl; j++) {
584 // choose the appropriate one (variable or target)
585 const VariableInfo& infoj = ( v_t == 0 ? Variable( j ) : Target(j) );
586 TString myVarj = infoj.GetInternalName();
587
588 Double_t rxmin = fVariableStats.at(fNumC-1).at( ( v_t*nvar )+ivar).fMin;
589 Double_t rxmax = fVariableStats.at(fNumC-1).at( ( v_t*nvar )+ivar).fMax;
590 Double_t rymin = fVariableStats.at(fNumC-1).at( ( v_t*nvar )+j).fMin;
591 Double_t rymax = fVariableStats.at(fNumC-1).at( ( v_t*nvar )+j).fMax;
592
593 // scatter plot
594 TH2F* h2 = new TH2F( TString::Format( "scat_%s_vs_%s_%s%s" , myVarj.Data(), myVari.Data(),
595 className.Data(), transfType.Data() ),
596 TString::Format( "%s versus %s (%s)%s", infoj.GetTitle(), info.GetTitle(),
597 className.Data(), transfType.Data() ),
598 nbins2D, rxmin , rxmax,
599 nbins2D, rymin , rymax );
600
601 h2->GetXaxis()->SetTitle( gTools().GetXTitleWithUnit( GetVariableAxisTitle( info ), info .GetUnit() ) );
602 h2->GetYaxis()->SetTitle( gTools().GetXTitleWithUnit( GetVariableAxisTitle( infoj ), infoj.GetUnit() ) );
603 mycorr.at(cls).at((var_tgt*nvar)+ivar).at((v_t*nvar)+j) = h2;
604
605 // profile plot
606 TProfile* p = new TProfile( TString::Format( "prof_%s_vs_%s_%s%s", myVarj.Data(),
607 myVari.Data(), className.Data(),
608 transfType.Data() ),
609 TString::Format( "profile %s versus %s (%s)%s",
610 infoj.GetTitle(), info.GetTitle(),
611 className.Data(), transfType.Data() ), nbins1D,
612 rxmin, rxmax );
613 // info.GetMin(), info.GetMax() );
614
615 p->GetXaxis()->SetTitle( gTools().GetXTitleWithUnit( GetVariableAxisTitle( info ), info .GetUnit() ) );
616 p->GetYaxis()->SetTitle( gTools().GetXTitleWithUnit( GetVariableAxisTitle( infoj ), infoj.GetUnit() ) );
617 myprof.at(cls).at((var_tgt*nvar)+ivar).at((v_t*nvar)+j) = p;
618 }
619 }
620 }
621 }
622 }
623 }
624
625 UInt_t nevts = events.size();
626
627 // compute correlation coefficient between target value and variables (regression only)
628 std::vector<Double_t> xregmean ( nvar+1, 0 );
629 std::vector<Double_t> x2regmean( nvar+1, 0 );
630 std::vector<Double_t> xCregmean( nvar+1, 0 );
631
632 // fill the histograms (this approach should be faster than individual projection
633 for (UInt_t ievt=0; ievt<nevts; ievt++) {
634
635 const Event* ev = events[ievt];
636
637 Float_t weight = ev->GetWeight();
638 Int_t cls = ev->GetClass();
639
640 // average correlation between first target and variables (so far only for single-target regression)
641 if (ntgt == 1) {
642 Float_t valr = ev->GetTarget(0);
643 xregmean[nvar] += valr;
644 x2regmean[nvar] += valr*valr;
645 for (UInt_t ivar=0; ivar<nvar; ivar++) {
646 Float_t vali = ev->GetValue(ivar);
647 xregmean[ivar] += vali;
648 x2regmean[ivar] += vali*vali;
649 xCregmean[ivar] += vali*valr;
650 }
651 }
652
653 // fill correlation histograms
654 for (UInt_t var_tgt = 0; var_tgt < 2; var_tgt++) { // create the histos first for the variables, then for the targets
655 UInt_t nloops = ( var_tgt == 0? nvar:ntgt ); // number of variables or number of targets
656 for (UInt_t ivar=0; ivar<nloops; ivar++) {
657 Float_t vali = ( var_tgt == 0 ? ev->GetValue(ivar) : ev->GetTarget(ivar) );
658
659 // variable histos
660 hVars.at(cls).at( ( var_tgt*nvar )+ivar)->Fill( vali, weight );
661
662 // correlation histos
664
665 for (UInt_t v_t = 0; v_t < 2; v_t++) {
666 UInt_t nl = ( v_t==0 ? nvar : ntgt );
667 UInt_t start = ( v_t==0 ? (var_tgt==0?ivar+1:0) : (var_tgt==0?nl:ivar+1) );
668 for (UInt_t j=start; j<nl; j++) {
669 Float_t valj = ( v_t == 0 ? ev->GetValue(j) : ev->GetTarget(j) );
670 mycorr.at(cls).at( ( var_tgt*nvar )+ivar).at( ( v_t*nvar )+j)->Fill( vali, valj, weight );
671 myprof.at(cls).at( ( var_tgt*nvar )+ivar).at( ( v_t*nvar )+j)->Fill( vali, valj, weight );
672 }
673 }
674 }
675 }
676 }
677 }
678
679 // correlation analysis for ranking (single-target regression only)
680 if (ntgt == 1) {
681 for (UInt_t ivar=0; ivar<=nvar; ivar++) {
682 xregmean[ivar] /= nevts;
683 x2regmean[ivar] = x2regmean[ivar]/nevts - xregmean[ivar]*xregmean[ivar];
684 }
685 for (UInt_t ivar=0; ivar<nvar; ivar++) {
686 xCregmean[ivar] = xCregmean[ivar]/nevts - xregmean[ivar]*xregmean[nvar];
687 xCregmean[ivar] /= TMath::Sqrt( x2regmean[ivar]*x2regmean[nvar] );
688 }
689
690 fRanking.push_back( new Ranking( GetName() + "Transformation", "|Correlation with target|" ) );
691 for (UInt_t ivar=0; ivar<nvar; ivar++) {
692 Double_t abscor = TMath::Abs( xCregmean[ivar] );
693 fRanking.back()->AddRank( Rank( fDataSetInfo.GetVariableInfo(ivar).GetLabel(), abscor ) );
694 }
695
696 if (nvar+ntgt <= (UInt_t)gConfig().GetVariablePlotting().fMaxNumOfAllowedVariablesForScatterPlots) {
697
698 // compute also mutual information (non-linear correlation measure)
699 fRanking.push_back( new Ranking( GetName() + "Transformation", "Mutual information" ) );
700 for (UInt_t ivar=0; ivar<nvar; ivar++) {
701 TH2F* h1 = mycorr.at(0).at( nvar ).at( ivar );
703 fRanking.back()->AddRank( Rank( fDataSetInfo.GetVariableInfo(ivar).GetLabel(), mi ) );
704 }
705
706 // compute correlation ratio (functional correlations measure)
707 fRanking.push_back( new Ranking( GetName() + "Transformation", "Correlation Ratio" ) );
708 for (UInt_t ivar=0; ivar<nvar; ivar++) {
709 TH2F* h2 = mycorr.at(0).at( nvar ).at( ivar );
710 Double_t cr = gTools().GetCorrelationRatio( *h2 );
711 fRanking.back()->AddRank( Rank( fDataSetInfo.GetVariableInfo(ivar).GetLabel(), cr ) );
712 }
713
714 // additionally compute correlation ratio from transposed histograms since correlation ratio is asymmetric
715 fRanking.push_back( new Ranking( GetName() + "Transformation", "Correlation Ratio (T)" ) );
716 for (UInt_t ivar=0; ivar<nvar; ivar++) {
717 TH2F* h2T = gTools().TransposeHist( *mycorr.at(0).at( nvar ).at( ivar ) );
718 Double_t cr = gTools().GetCorrelationRatio( *h2T );
719 fRanking.back()->AddRank( Rank( fDataSetInfo.GetVariableInfo(ivar).GetLabel(), cr ) );
720 delete h2T;
721 }
722 }
723 }
724 // computes ranking of input variables
725 // separation for 2-class classification
726 else if (fDataSetInfo.GetNClasses() == 2
727 && fDataSetInfo.GetClassInfo("Signal") != NULL
728 && fDataSetInfo.GetClassInfo("Background") != NULL
729 ) { // TODO: ugly hack.. adapt to new framework
730 fRanking.push_back( new Ranking( GetName() + "Transformation", "Separation" ) );
731 for (UInt_t i=0; i<nvar; i++) {
732 Double_t sep = gTools().GetSeparation( hVars.at(fDataSetInfo.GetClassInfo("Signal") ->GetNumber()).at(i),
733 hVars.at(fDataSetInfo.GetClassInfo("Background")->GetNumber()).at(i) );
734 fRanking.back()->AddRank( Rank( hVars.at(fDataSetInfo.GetClassInfo("Signal")->GetNumber()).at(i)->GetTitle(),
735 sep ) );
736 }
737 }
738
739 // for regression compute performance from correlation with target value
740
741 // write histograms
742
743 TDirectory* localDir = theDirectory;
744 if (theDirectory == 0) {
745 // create directory in root dir
746 fRootBaseDir->cd();
747 TString outputDir = TString("InputVariables");
748 TListIter trIt(&fTransformations);
749 while (VariableTransformBase *trf = (VariableTransformBase*) trIt())
750 outputDir += "_" + TString(trf->GetShortName());
751
752 TString uniqueOutputDir = outputDir;
753 Int_t counter = 0;
754 TObject* o = nullptr;
755 while( (o = fRootBaseDir->FindObject(uniqueOutputDir)) != nullptr ){
756 uniqueOutputDir = outputDir + TString::Format("_%d",counter);
757 Log() << kINFO << "A " << o->ClassName() << " with name " << o->GetName() << " already exists in "
758 << fRootBaseDir->GetPath() << ", I will try with "<<uniqueOutputDir<<"." << Endl;
759 ++counter;
760 }
761
762 // TObject* o = fRootBaseDir->FindObject(outputDir);
763 // if (o != 0) {
764 // Log() << kFATAL << "A " << o->ClassName() << " with name " << o->GetName() << " already exists in "
765 // << fRootBaseDir->GetPath() << "("<<outputDir<<")" << Endl;
766 // }
767 localDir = fRootBaseDir->mkdir( uniqueOutputDir );
768 localDir->cd();
769
770 Log() << kVERBOSE << "Create and switch to directory " << localDir->GetPath() << Endl;
771 }
772 else {
773 theDirectory->cd();
774 }
775
776 for (UInt_t i=0; i<nvar+ntgt; i++) {
777 for (Int_t cls = 0; cls < ncls; cls++) {
778 if (hVars.at(cls).at(i) != 0) {
779 hVars.at(cls).at(i)->Write();
780 hVars.at(cls).at(i)->SetDirectory(nullptr);
781 delete hVars.at(cls).at(i);
782 }
783 }
784 }
785
786 // correlation plots have dedicated directory
787 if (nvar+ntgt <= (UInt_t)gConfig().GetVariablePlotting().fMaxNumOfAllowedVariablesForScatterPlots) {
788
789 localDir = localDir->mkdir( "CorrelationPlots" );
790 localDir ->cd();
791 Log() << kDEBUG << "Create scatter and profile plots in target-file directory: " << Endl;
792 Log() << kDEBUG << localDir->GetPath() << Endl;
793
794
795 for (UInt_t i=0; i<nvar+ntgt; i++) {
796 for (UInt_t j=i+1; j<nvar+ntgt; j++) {
797 for (Int_t cls = 0; cls < ncls; cls++) {
798 if (mycorr.at(cls).at(i).at(j) != 0 ) {
799 mycorr.at(cls).at(i).at(j)->Write();
800 mycorr.at(cls).at(i).at(j)->SetDirectory(nullptr);
801 delete mycorr.at(cls).at(i).at(j);
802 }
803 if (myprof.at(cls).at(i).at(j) != 0) {
804 myprof.at(cls).at(i).at(j)->Write();
805 myprof.at(cls).at(i).at(j)->SetDirectory(nullptr);
806 delete myprof.at(cls).at(i).at(j);
807 }
808 }
809 }
810 }
811 }
812 if (theDirectory != 0 ) theDirectory->cd();
813 else fRootBaseDir->cd();
814}
815
816////////////////////////////////////////////////////////////////////////////////
817/// returns string for transformation
818
820{
821 VariableTransformBase* trf = ((VariableTransformBase*)GetTransformationList().Last());
822 if (!trf) return 0;
823 else return trf->GetTransformationStrings( fTransformationsReferenceClasses.back() );
824}
825
826////////////////////////////////////////////////////////////////////////////////
827/// returns string for transformation
828
830{
831 VariableTransformBase* trf = ((VariableTransformBase*)GetTransformationList().Last());
832 if (!trf) return 0;
833 else return trf->GetName();
834}
835
836////////////////////////////////////////////////////////////////////////////////
837/// write transformation to stream
838
840{
841 TListIter trIt(&fTransformations);
842 std::vector< Int_t >::const_iterator rClsIt = fTransformationsReferenceClasses.begin();
843
844 o << "NTransformtations " << fTransformations.GetSize() << std::endl << std::endl;
845
846 ClassInfo* ci;
847 UInt_t i = 1;
848 while (VariableTransformBase *trf = (VariableTransformBase*) trIt()) {
849 o << "#TR -*-*-*-*-*-*-* transformation " << i++ << ": " << trf->GetName() << " -*-*-*-*-*-*-*-" << std::endl;
850 trf->WriteTransformationToStream(o);
851 ci = fDataSetInfo.GetClassInfo( (*rClsIt) );
852 TString clsName;
853 if (ci == 0 ) clsName = "AllClasses";
854 else clsName = ci->GetName();
855 o << "ReferenceClass " << clsName << std::endl;
856 ++rClsIt;
857 }
858}
859
860
861////////////////////////////////////////////////////////////////////////////////
862/// XML node describing the transformation
863
865{
866 if(!parent) return;
867 void* trfs = gTools().AddChild(parent, "Transformations");
868 gTools().AddAttr( trfs, "NTransformations", fTransformations.GetSize() );
869 TListIter trIt(&fTransformations);
870 while (VariableTransformBase *trf = (VariableTransformBase*) trIt()) trf->AttachXMLTo(trfs);
871}
872
873////////////////////////////////////////////////////////////////////////////////
874
876{
877 Log() << kFATAL << "Read transformations not implemented" << Endl;
878 // TODO
879}
880
881////////////////////////////////////////////////////////////////////////////////
882
884{
885 void* ch = gTools().GetChild( trfsnode );
886 while(ch) {
887 Int_t idxCls = -1;
888 TString trfname;
889 gTools().ReadAttr(ch, "Name", trfname);
890
891 VariableTransformBase* newtrf = 0;
892
893 if (trfname == "Decorrelation" ) {
894 newtrf = new VariableDecorrTransform(fDataSetInfo);
895 }
896 else if (trfname == "PCA" ) {
897 newtrf = new VariablePCATransform(fDataSetInfo);
898 }
899 else if (trfname == "Gauss" ) {
900 newtrf = new VariableGaussTransform(fDataSetInfo);
901 }
902 else if (trfname == "Uniform" ) {
903 newtrf = new VariableGaussTransform(fDataSetInfo, "Uniform");
904 }
905 else if (trfname == "Normalize" ) {
906 newtrf = new VariableNormalizeTransform(fDataSetInfo);
907 }
908 else if (trfname == "Rearrange" ) {
909 newtrf = new VariableRearrangeTransform(fDataSetInfo);
910 }
911 else if (trfname != "None") {
912 }
913 else {
914 Log() << kFATAL << "<ReadFromXML> Variable transform '"
915 << trfname << "' unknown." << Endl;
916 }
917 newtrf->ReadFromXML( ch );
918 AddTransformation( newtrf, idxCls );
919 ch = gTools().GetNextChild(ch);
920 }
921}
922
923////////////////////////////////////////////////////////////////////////////////
924/// prints ranking of input variables
925
927{
928 //Log() << kINFO << " " << Endl;
929 Log() << kINFO << "Ranking input variables (method unspecific)..." << Endl;
930 std::vector<Ranking*>::const_iterator it = fRanking.begin();
931 for (; it != fRanking.end(); ++it) (*it)->Print();
932}
933
934////////////////////////////////////////////////////////////////////////////////
935
937{
938 try {
939 return fVariableStats.at(cls).at(ivar).fMean;
940 }
941 catch(...) {
942 try {
943 return fVariableStats.at(fNumC-1).at(ivar).fMean;
944 }
945 catch(...) {
946 Log() << kWARNING << "Inconsistent variable state when reading the mean value. " << Endl;
947 }
948 }
949 Log() << kWARNING << "Inconsistent variable state when reading the mean value. Value 0 given back" << Endl;
950 return 0;
951}
952
953////////////////////////////////////////////////////////////////////////////////
954
956{
957 try {
958 return fVariableStats.at(cls).at(ivar).fRMS;
959 }
960 catch(...) {
961 try {
962 return fVariableStats.at(fNumC-1).at(ivar).fRMS;
963 }
964 catch(...) {
965 Log() << kWARNING << "Inconsistent variable state when reading the RMS value. " << Endl;
966 }
967 }
968 Log() << kWARNING << "Inconsistent variable state when reading the RMS value. Value 0 given back" << Endl;
969 return 0;
970}
971
972////////////////////////////////////////////////////////////////////////////////
973
975{
976 try {
977 return fVariableStats.at(cls).at(ivar).fMin;
978 }
979 catch(...) {
980 try {
981 return fVariableStats.at(fNumC-1).at(ivar).fMin;
982 }
983 catch(...) {
984 Log() << kWARNING << "Inconsistent variable state when reading the minimum value. " << Endl;
985 }
986 }
987 Log() << kWARNING << "Inconsistent variable state when reading the minimum value. Value 0 given back" << Endl;
988 return 0;
989}
990
991////////////////////////////////////////////////////////////////////////////////
992
994{
995 try {
996 return fVariableStats.at(cls).at(ivar).fMax;
997 }
998 catch(...) {
999 try {
1000 return fVariableStats.at(fNumC-1).at(ivar).fMax;
1001 }
1002 catch(...) {
1003 Log() << kWARNING << "Inconsistent variable state when reading the maximum value. " << Endl;
1004 }
1005 }
1006 Log() << kWARNING << "Inconsistent variable state when reading the maximum value. Value 0 given back" << Endl;
1007 return 0;
1008}
#define h(i)
Definition RSha256.hxx:106
float Float_t
Definition RtypesCore.h:57
constexpr Bool_t kFALSE
Definition RtypesCore.h:101
const Bool_t kIterBackward
Definition TCollection.h:43
winID h TVirtualViewer3D TVirtualGLPainter p
Option_t Option_t TPoint TPoint const char x2
Option_t Option_t TPoint TPoint const char GetTextMagnitude GetFillStyle GetLineColor GetLineWidth GetMarkerStyle GetTextAlign GetTextColor GetTextSize void char Point_t Rectangle_t WindowAttributes_t Float_t Float_t Float_t Int_t Int_t UInt_t UInt_t Rectangle_t Int_t Int_t Window_t TString Int_t GCValues_t GetPrimarySelectionOwner GetDisplay GetScreen GetColormap GetNativeEvent const char const char dpyName wid window const char font_name cursor keysym reg const char only_if_exist regb h Point_t winding char text const char depth char const char Int_t count const char ColorStruct_t color const char Pixmap_t Pixmap_t PictureAttributes_t attr const char char ret_data h unsigned char height h Atom_t Int_t ULong_t ULong_t unsigned char prop_list Atom_t Atom_t Atom_t Time_t format
char name[80]
Definition TGX11.cxx:110
float xmin
float xmax
char * Form(const char *fmt,...)
Formats a string in a circular formatting buffer.
Definition TString.cxx:2489
Describe directory structure in memory.
Definition TDirectory.h:45
virtual const char * GetPath() const
Returns the full path of the directory.
virtual Bool_t cd()
Change current directory to "this" directory.
virtual TDirectory * mkdir(const char *name, const char *title="", Bool_t returnExistingDirectory=kFALSE)
Create a sub-directory "a" or a hierarchy of sub-directories "a/b/c/...".
1-D histogram with a float per channel (see TH1 documentation)
Definition TH1.h:621
TH1 is the base class of all histogram classes in ROOT.
Definition TH1.h:59
2-D histogram with a float per channel (see TH1 documentation)
Definition TH2.h:307
Iterator of linked list.
Definition TList.h:193
Class that contains all the information of a class.
Definition ClassInfo.h:49
Int_t fMaxNumOfAllowedVariablesForScatterPlots
Definition Config.h:109
VariablePlotting & GetVariablePlotting()
Definition Config.h:97
Class that contains all the data information.
Definition DataSetInfo.h:62
UInt_t GetNVariables() const
UInt_t GetNClasses() const
UInt_t GetNTargets() const
Float_t GetValue(UInt_t ivar) const
return value of i'th variable
Definition Event.cxx:236
Double_t GetWeight() const
return the event weight - depending on whether the flag IgnoreNegWeightsInTraining is or not.
Definition Event.cxx:389
UInt_t GetClass() const
Definition Event.h:86
Float_t GetTarget(UInt_t itgt) const
Definition Event.h:102
ostringstream derivative to redirect and format output
Definition MsgLogger.h:57
void SetSource(const std::string &source)
Definition MsgLogger.h:68
Ranking for variables in method (implementation)
Definition Ranking.h:48
Double_t GetSeparation(TH1 *S, TH1 *B) const
compute "separation" defined as
Definition Tools.cxx:121
Double_t GetMutualInformation(const TH2F &)
Mutual Information method for non-linear correlations estimates in 2D histogram Author: Moritz Backes...
Definition Tools.cxx:589
const TString & Color(const TString &)
human readable color strings
Definition Tools.cxx:828
Double_t GetCorrelationRatio(const TH2F &)
Compute Correlation Ratio of 2D histogram to estimate functional dependency between two variables Aut...
Definition Tools.cxx:620
void ReadAttr(void *node, const char *, T &value)
read attribute from xml
Definition Tools.h:329
void * GetChild(void *parent, const char *childname=nullptr)
get child node
Definition Tools.cxx:1150
void AddAttr(void *node, const char *, const T &value, Int_t precision=16)
add attribute to xml
Definition Tools.h:347
void * AddChild(void *parent, const char *childname, const char *content=nullptr, bool isRootNode=false)
add child node
Definition Tools.cxx:1124
TH2F * TransposeHist(const TH2F &)
Transpose quadratic histogram.
Definition Tools.cxx:657
void * GetNextChild(void *prevchild, const char *childname=nullptr)
XML helpers.
Definition Tools.cxx:1162
void AddXMLTo(void *parent=nullptr) const
XML node describing the transformation.
const char * GetNameOfLastTransform() const
returns string for transformation
void ReadFromStream(std::istream &istr)
TransformationHandler(DataSetInfo &, const TString &callerName)
constructor
const std::vector< Event * > * CalcTransformations(const std::vector< Event * > &, Bool_t createNewVector=kFALSE)
computation of transformation
const Event * Transform(const Event *) const
the transformation
Int_t fNumC
number of categories (#classes +1)
void AddStats(Int_t k, UInt_t ivar, Double_t mean, Double_t rms, Double_t min, Double_t max)
Caches calculated summary statistics of transformed variables.
Double_t GetRMS(Int_t ivar, Int_t cls=-1) const
Double_t GetMean(Int_t ivar, Int_t cls=-1) const
void SetCallerName(const TString &name)
std::vector< std::vector< TMVA::TransformationHandler::VariableStat > > fVariableStats
reference classes for the transformations
TString GetName() const
return transformation name
Double_t GetMin(Int_t ivar, Int_t cls=-1) const
void PrintVariableRanking() const
prints ranking of input variables
void MakeFunction(std::ostream &fout, const TString &fncName, Int_t part) const
create transformation function
void SetTransformationReferenceClass(Int_t cls)
overrides the setting for all classes! (this is put in basically for the likelihood-method) be carefu...
void PlotVariables(const std::vector< Event * > &events, TDirectory *theDirectory=nullptr)
create histograms from the input variables
void CalcStats(const std::vector< Event * > &events)
method to calculate minimum, maximum, mean, and RMS for all variables used in the MVA
std::vector< TString > * GetTransformationStringsOfLastTransform() const
returns string for transformation
TString GetVariableAxisTitle(const VariableInfo &info) const
incorporates transformation type into title axis (usually for histograms)
VariableTransformBase * AddTransformation(VariableTransformBase *, Int_t cls)
Double_t GetMax(Int_t ivar, Int_t cls=-1) const
const Event * InverseTransform(const Event *, Bool_t suppressIfNoTargets=true) const
void WriteToStream(std::ostream &o) const
write transformation to stream
@ kIdentity
Definition Types.h:115
Linear interpolation class.
Gaussian Transformation of input variables.
Class for type info of MVA input variable.
char GetVarType() const
const TString & GetInternalName() const
const TString & GetUnit() const
Linear interpolation class.
Rearrangement of input variables.
Linear interpolation class.
virtual void ReadFromXML(void *trfnode)=0
Read the input variables from the XML node.
virtual const char * GetName() const
Returns name of object.
virtual std::vector< TString > * GetTransformationStrings(Int_t cls) const
TODO --> adapt to variable,target,spectator selection default transformation output --> only indicate...
const char * GetName() const override
Returns name of object.
Definition TNamed.h:47
const char * GetTitle() const override
Returns title of object.
Definition TNamed.h:48
Mother of all ROOT objects.
Definition TObject.h:41
virtual const char * GetName() const
Returns name of object.
Definition TObject.cxx:439
virtual const char * ClassName() const
Returns name of class to which the object belongs.
Definition TObject.cxx:207
virtual TObject * FindObject(const char *name) const
Must be redefined in derived classes.
Definition TObject.cxx:403
virtual Int_t Write(const char *name=nullptr, Int_t option=0, Int_t bufsize=0)
Write this object to the current directory.
Definition TObject.cxx:880
virtual void Print(Option_t *option="") const
This method must be overridden when a class wants to print itself.
Definition TObject.cxx:636
Profile Histogram.
Definition TProfile.h:32
Basic string class.
Definition TString.h:139
const char * Data() const
Definition TString.h:376
static TString Format(const char *fmt,...)
Static method which formats a string using a printf style format descriptor and return a TString.
Definition TString.cxx:2378
Double_t x[n]
Definition legend1.C:17
TH1F * h1
Definition legend1.C:5
Config & gConfig()
Tools & gTools()
MsgLogger & Endl(MsgLogger &ml)
Definition MsgLogger.h:148
Bool_t IsNaN(Double_t x)
Definition TMath.h:892
Int_t Nint(T x)
Round to nearest integer. Rounds half integers to the nearest even integer.
Definition TMath.h:693
Short_t Max(Short_t a, Short_t b)
Returns the largest of a and b.
Definition TMathBase.h:250
Double_t Sqrt(Double_t x)
Returns the square root of x.
Definition TMath.h:662
Short_t Min(Short_t a, Short_t b)
Returns the smallest of a and b.
Definition TMathBase.h:198
Short_t Abs(Short_t d)
Returns the absolute value of parameter Short_t d.
Definition TMathBase.h:123