#!/usr/bin/env python
# coding: utf-8

# # tmva102_Testing
# This tutorial illustrates how you can test a trained BDT model using the fast
# tree inference engine offered by TMVA and external tools such as scikit-learn.
# 
# 
# 
# 
# **Author:** Stefan Wunsch  
# <i><small>This notebook tutorial was automatically generated with <a href= "https://github.com/root-project/root/blob/master/documentation/doxygen/converttonotebook.py">ROOTBOOK-izer</a> from the macro found in the ROOT repository  on Wednesday, April 17, 2024 at 11:22 AM.</small></i>

# In[ ]:


import ROOT
import pickle

from tmva100_DataPreparation import variables
from tmva101_Training import load_data


# Load data

# In[ ]:


x, y_true, w = load_data("test_signal.root", "test_background.root")


# Load trained model

# In[ ]:


File = "tmva101.root"
if (ROOT.gSystem.AccessPathName(File)) :
    ROOT.Info("tmva102_Testing.py", File+"does not exist")
    exit()

bdt = ROOT.TMVA.Experimental.RBDT[""]("myBDT", File)


# Make prediction

# In[ ]:


y_pred = bdt.Compute(x)


# Compute ROC using sklearn

# In[ ]:


from sklearn.metrics import roc_curve, auc
false_positive_rate, true_positive_rate, _ = roc_curve(y_true, y_pred, sample_weight=w)
score = auc(false_positive_rate, true_positive_rate)


# Plot ROC

# In[ ]:


c = ROOT.TCanvas("roc", "", 600, 600)
g = ROOT.TGraph(len(false_positive_rate), false_positive_rate, true_positive_rate)
g.SetTitle("AUC = {:.2f}".format(score))
g.SetLineWidth(3)
g.SetLineColor(ROOT.kRed)
g.Draw("AC")
g.GetXaxis().SetRangeUser(0, 1)
g.GetYaxis().SetRangeUser(0, 1)
g.GetXaxis().SetTitle("False-positive rate")
g.GetYaxis().SetTitle("True-positive rate")
c.Draw()