/*  In contrast to "updateReducedDeconvFisher.C", the 2D array variables are divided into
 *  1D array elements.
 *  Started 6/6/18 by John Russell.
 */

#include "AnitaTMVA.h"  //  Use this to take care of requisite TMVA includes.


void polarizedReducedDeconvFisher() {

	//  Open trees to be considered in TMVA.
	TFile signalFile("cutFiles/waisCutFiles/waisCutFileAll.root");
	TFile backgroundFile("cutFiles/decimatedThermalCutFiles/decimatedThermalCutFileAll.root");

	TTree * signalCutTree = (TTree *) signalFile.Get("cutTree");
	TTree * backgroundCutTree = (TTree *) backgroundFile.Get("cutTree");

	//  Create file in which to store TMVA results.
	TFile tmvaOutFile("tmvaFiles/polarizedReducedDeconvFisher.root", "RECREATE");

	//  Set up the TMVA factory.
	TMVA::Factory * factory = new TMVA::Factory("polarizedReducedFactory", & tmvaOutFile, "V");  //  The "V" is for verbose output.
	TMVA::DataLoader * dl = new TMVA::DataLoader("tmvaFiles");  // For interface with ROOT starting version 6.07/04.

	//  Variables to train on.
//	dl -> AddVariable("deconvWaveformCuts.waveformPeak[0]");
//	dl -> AddVariable("deconvWaveformCuts.waveformPeak[1]");
	dl -> AddVariable("deconvWaveformCuts.waveformPeakZScore[0]");
	dl -> AddVariable("deconvWaveformCuts.waveformPeakZScore[1]");
//	dl -> AddVariable("deconvWaveformCuts.analyticPeak[0]");
//	dl -> AddVariable("deconvWaveformCuts.analyticPeak[1]")
	dl -> AddVariable("deconvWaveformCuts.analyticPeakZScore[0]");
	dl -> AddVariable("deconvWaveformCuts.analyticPeakZScore[1]");
//	dl -> AddVariable("deconvWaveformCuts.snr[0]");
//	dl -> AddVariable("deconvWaveformCuts.snr[1]");
	dl -> AddVariable("deconvWaveformCuts.linearPolFrac[0]");
	dl -> AddVariable("deconvWaveformCuts.linearPolFrac[1]");
	dl -> AddVariable("deconvWaveformCuts.impulsivity[0]");
	dl -> AddVariable("deconvWaveformCuts.impulsivity[1]");
//	dl -> AddVariable("deconvWaveformCuts.autocorr5nsMean[0]");
//	dl -> AddVariable("deconvWaveformCuts.autocorrWaveform5nsRMS[0]");
//	dl -> AddVariable("deconvWaveformCuts.autocorrWaveform5nsValZScore[0]");
//	dl -> AddVariable("deconvWaveformCuts.autocorrAnalytic5nsRMS[0]");
//	dl -> AddVariable("deconvWaveformCuts.autocorrAnalytic5nsValZScore[0]");
//	dl -> AddVariable("deconvWaveformCuts.autocorrWaveformFullValZScore[0]");
//	dl -> AddVariable("deconvWaveformCuts.autocorrAnalyticFullValZScore[0]");
	dl -> AddVariable("deconvWaveformCuts.TIMdB[0]");
	dl -> AddVariable("deconvWaveformCuts.TIMdB[1]");
//	dl -> AddVariable("deconvWaveformCuts.corrPeak[0]");
//	dl -> AddVariable("deconvWaveformCuts.corrPeak[1]");
	dl -> AddVariable("deconvWaveformCuts.corrPeakZScore[0]");
	dl -> AddVariable("deconvWaveformCuts.corrPeakZScore[1]");
//	dl -> AddVariable("deconvWaveformCuts.analyticCorrPeak[0]");
//	dl -> AddVariable("deconvWaveformCuts.analyticCorrPeak[1]");
	dl -> AddVariable("deconvWaveformCuts.analyticCorrPeakZScore[0]");
	dl -> AddVariable("deconvWaveformCuts.analyticCorrPeakZScore[1]");
//	dl -> AddVariable("mapCuts.mapPeak[0]");
//	dl -> AddVariable("mapCuts.mapPeak[1]");
//	dl -> AddVariable("mapCuts.rectMapSNR[0]");
//	dl -> AddVariable("mapCuts.rectMapSNR[1]");
//	dl -> AddVariable("mapCuts.sphMapSNR[0]");
//	dl -> AddVariable("mapCuts.sphMapSNR[1]");
//	dl -> AddVariable("mapCuts.rectMapPeakZScore[0]");
//	dl -> AddVariable("mapCuts.rectMapPeakZScore[1]");
	dl -> AddVariable("mapCuts.sphMapPeakZScore[0]");
	dl -> AddVariable("mapCuts.sphMapPeakZScore[1]");
	dl -> AddVariable("deconvWaveformCuts.fishy[0]");
	dl -> AddVariable("deconvWaveformCuts.fishy[1]");

	//  Add spectator variables.
	dl -> AddSpectator("run");
	dl -> AddSpectator("eventNumber");
	dl -> AddSpectator("mapCuts.mapPeakNegTheta[0]");
	dl -> AddSpectator("mapCuts.mapPeakNegTheta[1]");

	//  Setting up signal and background (background) trees to be pointed towards Factory.
	dl -> AddSignalTree(signalCutTree);
	dl -> AddBackgroundTree(backgroundCutTree);

	// Choose Fisher linear discriminant analysis to perform testing.
	factory -> BookMethod(dl, TMVA::Types::kFisher, "deconvFisher", "");

	//  Collection of training, testing, and evaluating the MVA, then save the results to tree.
	factory -> TrainAllMethods();
	factory -> TestAllMethods();
	factory -> EvaluateAllMethods();
	tmvaOutFile.Write();
}