stt8_metrics.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## _Evaluation Metrics_\n",
    "\n",
    "_If **GNNBuilder** callback has been run during training, just load data from `dnn_processed/test` and extract `scores` and `y_pid ~ truth` and simply run the following metrics_."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys, os, glob, yaml"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pprint\n",
    "from tqdm import tqdm\n",
    "import trackml.dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchmetrics\n",
    "import pytorch_lightning as pl\n",
    "from torch_geometric.data import Data\n",
    "from torch_geometric.loader import DataLoader\n",
    "import itertools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = 'cuda' if torch.cuda.is_available() else 'cpu'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# append parent dir\n",
    "sys.path.append('..')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from src.metric_utils import compute_metrics, plot_metrics\n",
    "from src.metric_utils import plot_roc, plot_prc, plot_prc_thr, plot_epc, plot_epc_cut, plot_output"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": []
   },
   "source": [
    "### _Evaluation Definitions_\n",
    "\n",
    "Metrics to evaluate the GNN networks:\n",
    "\n",
    "- Accuracy/ACC = $TP+TN/TP+TN+FP+FN$\n",
    "- sensitivity, recall, hit rate, or true positive rate ($TPR = 1 - FNR$)\n",
    "- specificity, selectivity or true negative rate ($TNR = 1 - FPR$)\n",
    "- miss rate or false negative rate ($FNR = 1 - TPR$)\n",
    "- fall-out or false positive rate ($FPR = 1 - TNR$)\n",
    "- F1-score = $2 \\times (\\text{PPV} \\times \\text{TPR})/(\\text{PPV} + \\text{TPR})$\n",
    "- Efficiency/Recall/Sensitivity/Hit Rate: $TPR = TP/(TP+FN)$\n",
    "- Purity/Precision/Positive Predictive Value: $PPV = TP/(TP+FP$\n",
    "- AUC-ROC Curve $\\equiv$ FPR ($x-$axis) v.s. TPR ($y-$axis) plot\n",
    "- AUC-PRC Curve $\\equiv$ TPR ($x-$axis) v.s. PPV ($y-$axis) plot\n",
    "\n",
    "\n",
    "Use _`tn, fp, fn, tp = confusion_matrix(y_true, y_pred).ravel()`_ to directly access TN, FP, FN and TP using Scikit-learn."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": []
   },
   "source": [
    "### _Classifier Evaluation_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# DNN (Muons)\n",
    "# inputdir = \"run_all/dnn_processed_bn/pred\"\n",
    "# inputdir = \"run_all/dnn_processed_ln/pred\"\n",
    "\n",
    "# GNN (Muons)\n",
    "# inputdir = \"run_all/gnn_processed/pred\"\n",
    "\n",
    "# HypGNN (FWP + No Filtering)\n",
    "inputdir = \"run_all/fwp_gnn_processed/pred\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "outname = \"fwp\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of Files:  29703\n"
     ]
    }
   ],
   "source": [
    "test_files = sorted(glob.glob(os.path.join(inputdir, \"*\")))\n",
    "print(\"Number of Files: \", len(test_files))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Let's test a event\n",
    "data = torch.load(test_files[0], map_location=device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Data(x=[24, 3], pid=[24], layers=[24], event_file='/home/adeak977/current/3_deeptrkx/ctd2022/train_quick/event0000000000', hid=[24], pt=[24], vertex=[24, 3], charge=[24], pdgcode=[24], ptheta=[24], peta=[24], pphi=[24], modulewise_true_edges=[2, 22], layerwise_true_edges=[2, 25], edge_index=[2, 44], y_pid=[44], scores=[88])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### _Append Scores and Truths_\n",
    "- _Load all `truth` and `scores` from the `testset` from the `DNN` stage_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processed Batches:  1000\n",
      "Processed Batches:  2000\n",
      "Processed Batches:  3000\n",
      "Processed Batches:  4000\n",
      "Processed Batches:  5000\n",
      "Processed Batches:  6000\n",
      "Processed Batches:  7000\n",
      "Processed Batches:  8000\n",
      "Processed Batches:  9000\n",
      "Processed Batches:  10000\n",
      "Processed Batches:  11000\n",
      "Processed Batches:  12000\n",
      "Processed Batches:  13000\n",
      "Processed Batches:  14000\n",
      "Processed Batches:  15000\n",
      "Processed Batches:  16000\n",
      "Processed Batches:  17000\n",
      "Processed Batches:  18000\n",
      "Processed Batches:  19000\n",
      "Processed Batches:  20000\n",
      "Processed Batches:  21000\n",
      "Processed Batches:  22000\n",
      "Processed Batches:  23000\n",
      "Processed Batches:  24000\n",
      "Processed Batches:  25000\n",
      "Processed Batches:  26000\n",
      "Processed Batches:  27000\n",
      "Processed Batches:  28000\n",
      "Processed Batches:  29000\n"
     ]
    }
   ],
   "source": [
    "scoresl, truthsl = [], []\n",
    "\n",
    "for e in range(len(test_files)):\n",
    "\n",
    "    # read test events e.g. gnn_processed/test\n",
    "    graph = torch.load(test_files[e], map_location=device)\n",
    "    \n",
    "    # get truths and scores\n",
    "    truth = graph.y_pid\n",
    "    score = graph.scores\n",
    "    score = score[:truth.size(0)]\n",
    "\n",
    "    # logging\n",
    "    if e !=0 and (e)%1000==0:\n",
    "        print(\"Processed Batches: \", e)\n",
    "        \n",
    "    # append each batch\n",
    "    truthsl.append(truth)\n",
    "    scoresl.append(score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "scores = torch.cat(scoresl)\n",
    "truths = torch.cat(truthsl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save scores and truths as .npy files\n",
    "np.save(outname+\"_scores.npy\", scores.numpy())\n",
    "np.save(outname+\"_truths.npy\", truths.numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# torch to numpy\n",
    "scores = scores.numpy()\n",
    "truths = truths.numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### _Compute Metrics_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics = compute_metrics(scores,truths,threshold=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9824,0.9735,0.9871,0.9803\n"
     ]
    }
   ],
   "source": [
    "print(\"{:.4f},{:.4f},{:.4f},{:.4f}\".format(metrics.accuracy, metrics.precision, metrics.recall, metrics.f1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### _(a) - Plot Metrics_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plot_metrics(scores,truths, metrics, name=outname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ROC Curve\n",
    "# plot_roc(metrics, name=outname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# PR Curve\n",
    "# plot_prc(metrics, name=outname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Built from PRC Curve\n",
    "# plot_prc_thr(metrics, name=outname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# EP Curve from ROC\n",
    "plot_epc(metrics, name=outname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Built from ROC Curve\n",
    "plot_epc_cut(metrics, name=outname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Model output: True and False\n",
    "plot_output(scores, truths, threshold=0.5, name=outname)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": []
   },
   "source": [
    "### _(b) - S/B Suppression_\n",
    "\n",
    "Background rejection rate (1/FPR) is given as $1/\\epsilon_{bkg}$ where $\\epsilon_{bkg}$ is the fraction of fake edges that pass the classification requirement. Signal efficiency (TPR ~ Recall) ($\\epsilon_{sig}$) is defined as the number of true edges above a given classification score cut over the total number of true edges. What we have?\n",
    "\n",
    "- Signal Efficiency = $\\epsilon_{sig}$ = TPR ~ Recall \n",
    "- Background Rejection = $1 - \\epsilon_{bkg}$ ???\n",
    "- Background Rejection Rate = $1/\\epsilon_{bkg}$ = 1/FPR\n",
    "\n",
    "\n",
    "First apply a edge score cut to binarized the `scores`, we will call it `preds`. The count number of false or true edges that pass this cut. Then calculated background rejection rate and signal efficiency. For making a plot one can do calculations in batch by batch mode on the test dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "threshold = 0.7"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- _recall/tpr and fpr_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "preds, targets = scores, truths\n",
    "y_pred, y_true = (preds > threshold), (targets > threshold)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Confusion Matrix\n",
    "tn, fp, fn, tp = confusion_matrix(y_true, y_pred).ravel()\n",
    "\n",
    "# Find Recal/TPR and FPR\n",
    "tpr = tp / (tp + fn)\n",
    "fpr = fp / (fp + tn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.975793902654237, 0.011099999950762952)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tpr, fpr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **_edge efficiency vs edge purity_**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "edge efficiency: 0.9758\n",
      "    edge purity: 0.9889\n"
     ]
    }
   ],
   "source": [
    "# edge efficiency (TPR), purity (1 - FPR)\n",
    "print(\"edge efficiency: {:.4f}\\n    edge purity: {:.4f}\".format(tpr, (1-fpr)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **_signal vs background rejection rate_**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Signal efficiency: 0.9758\n",
      "Misidentification rate: 0.0111\n",
      "         bkg rejection: 90.0901\n"
     ]
    }
   ],
   "source": [
    "# Signal efficiency (TPR), Bkg efficiency (FPR), Bkg Rejection (1/FPR)\n",
    "print(\"     Signal efficiency: {:.4f}\\nMisidentification rate: {:.4f}\\n         bkg rejection: {:.4f}\".format(tpr, fpr, (1/fpr)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- _plotting signal vs bkg rejection_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Metrics with Threshold\n",
    "metrics = compute_metrics(scores,truths,threshold)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_1817619/712045607.py:3: RuntimeWarning: divide by zero encountered in true_divide\n",
      "  bkg_rejection = 1/metrics.roc_fpr\n"
     ]
    }
   ],
   "source": [
    "# ROC Curve\n",
    "sig = metrics.roc_tpr\n",
    "bkg_rejection = 1/metrics.roc_fpr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "# cut off eff < 0.2 or 0.5\n",
    "sig_mask = sig > 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(8,6))\n",
    "ax.plot(sig[sig_mask], bkg_rejection[sig_mask], label=\"Interaction Network\", color=\"blue\")\n",
    "ax.plot(tpr, 1/fpr, marker=\"o\", markersize=10, markeredgecolor=\"k\", markerfacecolor=\"k\", label=\"Edge Score Cut = 0.5\", color=\"k\")\n",
    "\n",
    "# Axes Params\n",
    "ax.set_xlabel(\"Signal Efficiency\", fontsize=16)\n",
    "ax.set_ylabel(\"Background Rejection\", fontsize=16)\n",
    "ax.set_yscale('log')\n",
    "ax.tick_params(axis='both', which='major', labelsize=12)\n",
    "ax.tick_params(axis='both', which='minor', labelsize=12)\n",
    "ax.grid(True)\n",
    "ax.legend(fontsize=14, loc='upper right')\n",
    "    \n",
    "# Figure Params\n",
    "fig.tight_layout()\n",
    "fig.savefig(outname+\"_SB.pdf\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### _(c) - Visualize Model Output_"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "from src.drawing import detector_layout\n",
    "from src.utils_math import polar_to_cartesian"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "filter_files = [18,106,109,113,120,122,133,139,147,152,153,158,159,164,1012,1022,1030,1031,1040,1894,1892,1880,1877,\n",
    "                1872,1860,1857,1828,1827,1817,1816,1816,1807,1804,1767,1761,1751,1750,1749,1743,1734,1722,1721]"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "e = filter_files[5191]"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "e"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "# load graph\n",
    "graph = torch.load(test_files[5191], map_location=device)\n",
    "\n",
    "# get truths and scores\n",
    "truth = graph.y_pid\n",
    "scores = graph.scores[:truth.size(0)]\n",
    "edges = graph.edge_index\n",
    "eid = int(graph.event_file[-10:])"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "eid"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "truth.shape, scores.shape, edges.shape"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "preds, labels = scores.numpy(), truth.numpy()"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "preds.shape, labels.shape"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "def draw_sample_xy(graph, cut=0.5, figsize=(9,9)):\n",
    "    \"\"\"\"Draw Sample with True and False Edges\"\"\"\n",
    "    \n",
    "    # coordinate transformation\n",
    "    x = graph.x.detach().numpy()\n",
    "    r, phi, ir = x.T\n",
    "    x, y = polar_to_cartesian(r, phi)\n",
    "    \n",
    "    \n",
    "    truth = graph.y_pid\n",
    "    scores = graph.scores[:truth.size(0)]\n",
    "    edges = graph.edge_index\n",
    "    eid = int(graph.event_file[-5:])\n",
    "    preds, labels = scores.numpy(), truth.numpy()\n",
    "    \n",
    "    \n",
    "    # detector layout\n",
    "    fig, ax = detector_layout(figsize=figsize)\n",
    "    \n",
    "    # True Event\n",
    "    pids = np.unique(graph.pid)    \n",
    "    for pid in pids:\n",
    "        idx = graph.pid == pid\n",
    "        ax.plot(x[idx], y[idx], 'k-', linewidth=1.5)\n",
    "        ax.scatter(x[idx], y[idx], label='particle_id: {}'.format(int(pid)))\n",
    "    \n",
    "    \n",
    "    # Draw the segments\n",
    "    for j in range(labels.shape[0]):\n",
    "        \n",
    "        ptx1 = x[edges[0,j]]\n",
    "        ptx2 = x[edges[1,j]]\n",
    "        pty1 = y[edges[0,j]]\n",
    "        pty2 = y[edges[1,j]]\n",
    "        \n",
    "        # False Negatives\n",
    "        if preds[j] < cut and labels[j] > cut:\n",
    "            # ax.plot([x[edges[0,j]], x[edges[1,j]]], [y[edges[0,j]], y[edges[1,j]]], '--', c='b')\n",
    "            ax.plot([ptx1, ptx2], [pty1, pty2], '--', color='b', lw=1, alpha=0.9)\n",
    "\n",
    "        # False Positives\n",
    "        if preds[j] > cut and labels[j] < cut:\n",
    "            # ax.plot([x[edges[0,j]], x[edges[1,j]]], [y[edges[0,j]], y[edges[1,j]]], '-', c='r', alpha=preds[j])\n",
    "            ax.plot([ptx1, ptx2], [pty1, pty2], '-', color='r', lw=1, alpha=0.15)\n",
    "\n",
    "        # True Positives\n",
    "        if preds[j] > cut and labels[j] > cut:\n",
    "            # ax.plot([x[edges[0,j]], x[edges[1,j]]], [y[edges[0,j]], y[edges[1,j]]], '-', c='k', alpha=preds[j])\n",
    "            ax.plot([ptx1, ptx2], [pty1, pty2], '-', color='k', lw=1, alpha=0.3)\n",
    "\n",
    "    fig.tight_layout()\n",
    "    fig.savefig(\"ambiguous1_{}.pdf\".format(eid))"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "# draw_sample_xy(graph, cut=0.7);"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "def draw_sample_xy(graph, lower_cut=0.6, upper_cut=0.8, figsize=(8,8)):\n",
    "    \"\"\"\"Draw Sample with True and False Edges\"\"\"\n",
    "    \n",
    "    # coordinate transformation\n",
    "    x = graph.x.detach().numpy()\n",
    "    truth = graph.y_pid\n",
    "    scores = graph.scores[:truth.size(0)]\n",
    "    edges = graph.edge_index\n",
    "    eid = int(graph.event_file[-5:])\n",
    "    pids = np.unique(graph.pid)\n",
    "    \n",
    "    preds, labels = scores.numpy(), truth.numpy()\n",
    "    \n",
    "    \n",
    "    # detector layout\n",
    "    fig, ax = detector_layout(figsize=figsize)\n",
    "    \n",
    "    # True Event\n",
    "    r, phi, ir = x.T\n",
    "    x, y = polar_to_cartesian(r, phi)\n",
    "    \n",
    "    for pid in pids:\n",
    "        idx = graph.pid == pid\n",
    "        ax.plot(x[idx], y[idx], linewidth=1)\n",
    "        ax.scatter(x[idx], y[idx], label='particle_id: {}'.format(int(pid)))\n",
    "    \n",
    "    \n",
    "    # Draw the segments\n",
    "    for j in range(labels.shape[0]):\n",
    "        \n",
    "        ptx1 = x[edges[0,j]]\n",
    "        ptx2 = x[edges[1,j]]\n",
    "        pty1 = y[edges[0,j]]\n",
    "        pty2 = y[edges[1,j]]\n",
    "        \n",
    "        # False Negatives\n",
    "        if preds[j] < lower_cut and labels[j] > upper_cut:\n",
    "            # ax.plot([x[edges[0,j]], x[edges[1,j]]], [y[edges[0,j]], y[edges[1,j]]], '--', c='b')\n",
    "            ax.plot([ptx1, ptx2], [pty1, pty2], '--', color='b', lw=1, alpha=0.9)\n",
    "\n",
    "        # False Positives\n",
    "        if preds[j] > upper_cut and labels[j] < lower_cut:\n",
    "            # ax.plot([x[edges[0,j]], x[edges[1,j]]], [y[edges[0,j]], y[edges[1,j]]], '-', c='r', alpha=preds[j])\n",
    "            ax.plot([ptx1, ptx2], [pty1, pty2], '-', color='r', lw=1, alpha=0.15)\n",
    "\n",
    "        # True Positives\n",
    "        if preds[j] > lower_cut and labels[j] > upper_cut:\n",
    "            # ax.plot([x[edges[0,j]], x[edges[1,j]]], [y[edges[0,j]], y[edges[1,j]]], '-', c='k', alpha=preds[j])\n",
    "            ax.plot([ptx1, ptx2], [pty1, pty2], '-', color='k', lw=1, alpha=0.3)\n",
    "    \n",
    "    ax.set_title('Azimuthal View of STT, EventID # {}'.format(eid))\n",
    "    fig.tight_layout()\n",
    "    fig.savefig(\"ambiguous_{}.png\".format(eid))"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "draw_sample_xy(graph, lower_cut=0.7, upper_cut=0.8)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "for e in filter_files:\n",
    "    # read test events e.g. gnn_processed/test\n",
    "    graph = torch.load(test_files[e], map_location=device)\n",
    "    draw_sample_xy(graph, lower_cut=0.65, upper_cut=0.75);"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "# extract hit information\n",
    "r, phi, ir = graph.x.T\n",
    "ir = ir.detach().numpy()*100\n",
    "x, y = polar_to_cartesian(r.detach().numpy(), phi.detach().numpy())"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "# plot true event\n",
    "fig, ax = detector_layout(figsize=(10,10))\n",
    "pids = np.unique(graph.pid)\n",
    "\n",
    "# True Event\n",
    "for pid in pids:\n",
    "    idx = graph.pid == pid\n",
    "    ax.plot(x[idx], y[idx], '-', linewidth=1.5)\n",
    "    ax.scatter(x[idx], y[idx], label='particle_id: {}'.format(int(pid)))\n",
    "\n",
    "ax.set_title('Azimuthal View of STT, EventID # {}'.format(eid))\n",
    "ax.legend(fontsize=10, loc='best')\n",
    "fig.tight_layout()\n",
    "# fig.savefig(\"true_track.png\")"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}