/**********************************************************************

* Copyright 2014 Kalanand Mishra *

* *

* Code to predict optimal car insurance policy. See for details: *

https://www.kaggle.com/c/allstate-purchase-prediction-challenge *

* *

* Uses ROOT - a freely distributed software: http://root.cern.ch *

**********************************************************************/

#include "PrintResult.C"

void Analyze() {

// --- convert the csv to ROOT files if needed ------

makeTree("train");

makeTree("test");

// --- prepare tmva training ------------

char* trainFile = "train.root";

char* testFile = "test.root";

char* tree = "ntuple";

char* mycuts = "abs(record_type-1)<0.1"; // signal

char* mycutb = "abs(record_type)<0.1"; // background

char* inputVars[] = { "state_fips", "location", "group_size", "homeowner", "car_age", "car_value_num", "risk_factor", "age_oldest", "age_youngest", "married_couple", "C_previous", "duration_previous", "A", "B", "C", "D", "E", "F", "G", "cost" };

int size = sizeof(inputVars)/sizeof(inputVars[0]);

// TMVAClassification(trainFile, tree, mycuts, mycutb, inputVars, size);

// --- compute output and print result in a CSV file ------------

PrintResult(testFile, inputVars, size, "output.csv");

}

// Read data from an ascii file and create a root file with an ntuple.

void makeTree(char* txtFile) {

char state[20];

char car_value[20];

float state_fips;

float car_value_num;

TFile *f = new TFile((TString(txtFile)+TString(".root")).Data(),"RECREATE");

TTree *tree = new TTree("ntuple","data from ascii file");

const char* branchDescriptor = "customer_ID/I:shopping_pt/F:record_type:day:time/C:state/C:location/F:group_size:homeowner:car_age:car_value/C:risk_factor/F:age_oldest:age_youngest:married_couple:C_previous:duration_previous:A:B:C:D:E:F:G:cost";

Long64_t nlines = tree->ReadFile((TString(txtFile)+TString(".csv")).Data(), branchDescriptor);

printf(" found %d points\n",nlines);

tree->SetBranchAddress("state", state);

tree->SetBranchAddress("car_value", car_value);

TBranch *fips_branch = tree->Branch("state_fips",

&state_fips,"state_fips/F");

TBranch *car_branch = tree->Branch("car_value_num",

&car_value_num,"car_value_num/F");

Long64_t nentries = tree->GetEntries();

for (Int_t i = 0; i < nentries; i++) {

tree->GetEntry(i);

state_fips = (float) GetStateFIPS(state);

fips_branch->Fill();

car_value_num = (float) GetCarValNum(car_value);

car_branch->Fill();

}

tree->Write();

f->Close();

}

/**********************************************************************************

* Training and testing of the TMVA classifiers. *

* As input data is used a toy-MC sample consisting of four Gaussian-distributed *

* and linearly correlated input variables. *

**********************************************************************************/

void TMVAClassification(char* trainFile, char* tree,

char* mycuts, char* mycutb, char* inputVars[], int size)

{

// this loads the library

TMVA::Tools::Instance();

// Create a new root output file.

TFile* outputFile = TFile::Open( "TMVA.root", "RECREATE" );

// Create the factory object.

TMVA::Factory *factory = new TMVA::Factory( "TMVAClassification", outputFile,

"!V:!Silent:Color:DrawProgressBar:Transformations=I;D;P;G,D" );

// ---------- input variables

for (int ivar = 0; ivar < size; ++ivar) {

factory->AddVariable(inputVars[ivar], 'F');

}

// read training and test data

TFile *input = TFile::Open( trainFile);

TTree *signal = (TTree*)input->Get(tree);

TTree *background = (TTree*)input->Get(tree);

// global event weights per tree

Double_t signalWeight = 1.0;

Double_t backgroundWeight = 1.0;

// ====== register trees ====================================================

// you can add an arbitrary number of signal or background trees

factory->AddSignalTree ( signal, signalWeight );

factory->AddBackgroundTree( background, backgroundWeight );

// tell the factory to use all remaining events in the trees after training for testing:

factory->PrepareTrainingAndTestTree( TCut(mycuts), TCut(mycutb),

"nTrain_Signal=0:nTrain_Background=0:SplitMode=Random:NormMode=NumEvents:!V" );

// If no numbers of events are given, half of the events in the tree are used for training, and

// the other half for testing:

// ---- Use BDT: Adaptive Boost

factory->BookMethod( TMVA::Types::kBDT, "BDT",

"!H:!V:NTrees=400:nEventsMin=400:MaxDepth=3:BoostType=AdaBoost:SeparationType=GiniIndex:nCuts=20:PruneMethod=NoPruning" );

// ---- Train MVAs using the set of training events

factory->TrainAllMethods();

// ---- Evaluate all MVAs using the set of test events

factory->TestAllMethods();

// ----- Evaluate and compare performance of all configured MVAs

factory->EvaluateAllMethods();

// Save the output

outputFile->Close();

std::cout << "==> TMVAClassification is done!" << std::endl;

delete factory;

}

int GetStateFIPS(char* state) {

if(strcmp(state,"AL")==0) return(1);

if(strcmp(state,"AK")==0) return(2);

if(strcmp(state,"AZ")==0) return(4);

if(strcmp(state,"AR")==0) return(5);

if(strcmp(state,"CA")==0) return(6);

if(strcmp(state,"CO")==0) return(8);

if(strcmp(state,"CT")==0) return(9);

if(strcmp(state,"DE")==0) return(10);

if(strcmp(state,"FL")==0) return(12);

if(strcmp(state,"GA")==0) return(13);

if(strcmp(state,"HI")==0) return(15);

if(strcmp(state,"ID")==0) return(16);

if(strcmp(state,"IL")==0) return(17);

if(strcmp(state,"IN")==0) return(18);

if(strcmp(state,"IA")==0) return(19);

if(strcmp(state,"KS")==0) return(20);

if(strcmp(state,"KY")==0) return(21);

if(strcmp(state,"LA")==0) return(22);

if(strcmp(state,"ME")==0) return(23);

if(strcmp(state,"MD")==0) return(24);

if(strcmp(state,"MA")==0) return(25);

if(strcmp(state,"MI")==0) return(26);

if(strcmp(state,"MN")==0) return(27);

if(strcmp(state,"MS")==0) return(28);

if(strcmp(state,"MO")==0) return(29);

if(strcmp(state,"MT")==0) return(30);

if(strcmp(state,"NE")==0) return(31);

if(strcmp(state,"NV")==0) return(32);

if(strcmp(state,"NH")==0) return(33);

if(strcmp(state,"NJ")==0) return(34);

if(strcmp(state,"NM")==0) return(35);

if(strcmp(state,"NY")==0) return(36);

if(strcmp(state,"NC")==0) return(37);

if(strcmp(state,"ND")==0) return(38);

if(strcmp(state,"OH")==0) return(39);

if(strcmp(state,"OK")==0) return(40);

if(strcmp(state,"OR")==0) return(41);

if(strcmp(state,"PA")==0) return(42);

if(strcmp(state,"RI")==0) return(44);

if(strcmp(state,"SC")==0) return(45);

if(strcmp(state,"SD")==0) return(46);

if(strcmp(state,"TN")==0) return(47);

if(strcmp(state,"TX")==0) return(48);

if(strcmp(state,"UT")==0) return(49);

if(strcmp(state,"VT")==0) return(50);

if(strcmp(state,"VA")==0) return(51);

if(strcmp(state,"WA")==0) return(53);

if(strcmp(state,"WV")==0) return(54);

if(strcmp(state,"WI")==0) return(55);

if(strcmp(state,"WY")==0) return(56);

if(strcmp(state,"AS")==0) return(60);

if(strcmp(state,"GU")==0) return(66);

if(strcmp(state,"MP")==0) return(69);

if(strcmp(state,"PR")==0) return(72);

if(strcmp(state,"VI")==0) return(78);

return 100;

}

int GetCarValNum(char* car_value) {

if(strcmp(car_value,"a")==0) return 1;

if(strcmp(car_value,"b")==0) return 2;

if(strcmp(car_value,"c")==0) return 3;

if(strcmp(car_value,"d")==0) return 4;

if(strcmp(car_value,"e")==0) return 5;

if(strcmp(car_value,"f")==0) return 6;

if(strcmp(car_value,"g")==0) return 7;

if(strcmp(car_value,"h")==0) return 8;

if(strcmp(car_value,"i")==0) return 9;

return 0;

}