evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "proprietary-plymouth",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import roc_curve, precision_recall_curve\n",
    "import seaborn as sns\n",
    "from pathlib import Path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "figured-spending",
   "metadata": {},
   "outputs": [],
   "source": [
    "sns.set(font_scale=1.5)\n",
    "sns.set_style(\"ticks\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "foreign-august",
   "metadata": {},
   "outputs": [],
   "source": [
    "# results_path = \"stead_phasenet/version_2\"\n",
    "# results_path = \"stead_eqtransformer/version_0\"\n",
    "# results_path = \"ethz_eqtransformer/version_0\"\n",
    "# results_path = \"ethz_phasenet/version_0\"\n",
    "# results_path = \"stead_gpd/version_1\"\n",
    "# results_path = \"stead_cred/version_0\"\n",
    "results_path = \"instance_gpdpick_0.01/version_0\"\n",
    "\n",
    "split = \"dev\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "subtle-found",
   "metadata": {},
   "source": [
    "# Task 1 - Event detection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "sunset-norwegian",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = pd.read_csv(f\"pred/{results_path}/{split}_task1.csv\")\n",
    "\n",
    "pred[\"trace_type_bin\"] = pred[\"trace_type\"] == \"earthquake\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "regulation-conflict",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'true positive rate')"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fpr, tpr, _ = roc_curve(pred[\"trace_type_bin\"], pred[\"score_detection\"])\n",
    "\n",
    "fig = plt.figure(figsize=(10, 10))\n",
    "ax = fig.add_subplot(111)\n",
    "ax.plot(fpr, tpr)\n",
    "ax.set_aspect(\"equal\")\n",
    "\n",
    "ax.set_xlabel(\"false positive rate\")\n",
    "ax.set_ylabel(\"true positive rate\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "invisible-brave",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimal F1: 0.9646649260226284\n",
      "Optimal F1 threshold: 0.48392799999999997\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "prec, recall, thr = precision_recall_curve(pred[\"trace_type_bin\"], pred[\"score_detection\"])\n",
    "\n",
    "f1 = 2 * prec * recall / (prec + recall)\n",
    "f1_threshold = thr[np.nanargmax(f1)]\n",
    "best_f1 = np.max(f1)\n",
    "\n",
    "fig = plt.figure(figsize=(10, 10))\n",
    "ax = fig.add_subplot(111)\n",
    "ax.plot(prec, recall)\n",
    "ax.set_aspect(\"equal\")\n",
    "\n",
    "ax.set_xlabel(\"Precision\")\n",
    "ax.set_ylabel(\"Recall\")\n",
    "ax.set_xlim(ax.get_ylim())\n",
    "\n",
    "print(\"Optimal F1:\", best_f1)\n",
    "print(\"Optimal F1 threshold:\", f1_threshold)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "surgical-aviation",
   "metadata": {},
   "source": [
    "# Task 2 - Phase identification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "expensive-wrestling",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = pd.read_csv(f\"pred/{results_path}/{split}_task23.csv\")\n",
    "pred[\"phase_label_bin\"] = pred[\"phase_label\"] == \"P\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "introductory-fellowship",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'true positive rate')"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fpr, tpr, _ = roc_curve(pred[\"phase_label_bin\"], pred[\"score_p_or_s\"])\n",
    "\n",
    "fig = plt.figure(figsize=(10, 10))\n",
    "ax = fig.add_subplot(111)\n",
    "ax.plot(fpr, tpr)\n",
    "ax.set_aspect(\"equal\")\n",
    "\n",
    "ax.set_xlabel(\"false positive rate\")\n",
    "ax.set_ylabel(\"true positive rate\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "dense-supervision",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimal F1: 0.9765025885517473\n",
      "Optimal F1 threshold: 0.9023040000000001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "prec, recall, thr = precision_recall_curve(pred[\"phase_label_bin\"], pred[\"score_p_or_s\"])\n",
    "\n",
    "f1 = 2 * prec * recall / (prec + recall)\n",
    "f1_threshold = thr[np.nanargmax(f1)]\n",
    "best_f1 = np.max(f1)\n",
    "\n",
    "fig = plt.figure(figsize=(10, 10))\n",
    "ax = fig.add_subplot(111)\n",
    "ax.plot(prec, recall)\n",
    "ax.set_aspect(\"equal\")\n",
    "\n",
    "ax.set_xlabel(\"Precision\")\n",
    "ax.set_ylabel(\"Recall\")\n",
    "ax.set_xlim(ax.get_ylim())\n",
    "\n",
    "print(\"Optimal F1:\", best_f1)\n",
    "print(\"Optimal F1 threshold:\", f1_threshold)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "secure-trance",
   "metadata": {},
   "source": [
    "# Task 3 - Onset determination"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "minimal-deployment",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = pd.read_csv(f\"pred/{results_path}/{split}_task23.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "shaped-twelve",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(15, 10))\n",
    "axs = fig.subplots(1, 2)\n",
    "\n",
    "for i, phase in enumerate([\"P\", \"S\"]):\n",
    "    ax = axs[i]\n",
    "    ax.set_title(f\"{phase} arrivals\")\n",
    "    pred_phase = pred[pred[\"phase_label\"] == phase]\n",
    "    pred_col = f\"{phase.lower()}_sample_pred\"\n",
    "    \n",
    "    diff = (pred_phase[pred_col] - pred_phase[\"phase_onset\"]) / pred_phase[\"sampling_rate\"]\n",
    "    \n",
    "    bins = np.linspace(-2.5, 2.5, 50)\n",
    "    ax.hist(diff, bins=bins)\n",
    "    ax.set_xlabel(\"$t_{pred} - t_{true}$\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "sufficient-stomach",
   "metadata": {},
   "source": [
    "# Systematic evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "norwegian-gregory",
   "metadata": {},
   "outputs": [],
   "source": [
    "results = pd.read_csv(\"results.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "white-peace",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>experiment</th>\n",
       "      <th>data</th>\n",
       "      <th>model</th>\n",
       "      <th>lr</th>\n",
       "      <th>version</th>\n",
       "      <th>dev_phase_precision</th>\n",
       "      <th>dev_phase_recall</th>\n",
       "      <th>dev_phase_f1</th>\n",
       "      <th>phase_threshold</th>\n",
       "      <th>test_phase_precision</th>\n",
       "      <th>...</th>\n",
       "      <th>test_S_mean_s</th>\n",
       "      <th>test_S_std_s</th>\n",
       "      <th>test_S_mae_s</th>\n",
       "      <th>dev_det_precision</th>\n",
       "      <th>dev_det_recall</th>\n",
       "      <th>dev_det_f1</th>\n",
       "      <th>det_threshold</th>\n",
       "      <th>test_det_precision</th>\n",
       "      <th>test_det_recall</th>\n",
       "      <th>test_det_f1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ethz_basicphaseae_0.0001</td>\n",
       "      <td>ethz</td>\n",
       "      <td>basicphaseae</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>0</td>\n",
       "      <td>0.818765</td>\n",
       "      <td>0.891444</td>\n",
       "      <td>0.853560</td>\n",
       "      <td>7.936640</td>\n",
       "      <td>0.751454</td>\n",
       "      <td>...</td>\n",
       "      <td>0.201656</td>\n",
       "      <td>1.309772</td>\n",
       "      <td>0.630289</td>\n",
       "      <td>0.932951</td>\n",
       "      <td>0.762390</td>\n",
       "      <td>0.839091</td>\n",
       "      <td>0.316741</td>\n",
       "      <td>0.921642</td>\n",
       "      <td>0.702241</td>\n",
       "      <td>0.797120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ethz_basicphaseae</td>\n",
       "      <td>ethz</td>\n",
       "      <td>basicphaseae</td>\n",
       "      <td>0.0010</td>\n",
       "      <td>0</td>\n",
       "      <td>0.920499</td>\n",
       "      <td>0.967113</td>\n",
       "      <td>0.943230</td>\n",
       "      <td>2.620540</td>\n",
       "      <td>0.868497</td>\n",
       "      <td>...</td>\n",
       "      <td>0.065660</td>\n",
       "      <td>1.241713</td>\n",
       "      <td>0.533804</td>\n",
       "      <td>0.956396</td>\n",
       "      <td>0.911894</td>\n",
       "      <td>0.933615</td>\n",
       "      <td>0.296525</td>\n",
       "      <td>0.939023</td>\n",
       "      <td>0.885646</td>\n",
       "      <td>0.911554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ethz_basicphaseae_0.01</td>\n",
       "      <td>ethz</td>\n",
       "      <td>basicphaseae</td>\n",
       "      <td>0.0100</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ethz_cred_0.0001</td>\n",
       "      <td>ethz</td>\n",
       "      <td>cred</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.903446</td>\n",
       "      <td>0.945485</td>\n",
       "      <td>0.923988</td>\n",
       "      <td>0.077826</td>\n",
       "      <td>0.883105</td>\n",
       "      <td>0.935241</td>\n",
       "      <td>0.908426</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ethz_cred</td>\n",
       "      <td>ethz</td>\n",
       "      <td>cred</td>\n",
       "      <td>0.0010</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.920923</td>\n",
       "      <td>0.923458</td>\n",
       "      <td>0.922189</td>\n",
       "      <td>0.061431</td>\n",
       "      <td>0.890877</td>\n",
       "      <td>0.911779</td>\n",
       "      <td>0.901207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>stead_gpdpick</td>\n",
       "      <td>stead</td>\n",
       "      <td>gpdpick</td>\n",
       "      <td>0.0010</td>\n",
       "      <td>0</td>\n",
       "      <td>0.994568</td>\n",
       "      <td>0.991264</td>\n",
       "      <td>0.992913</td>\n",
       "      <td>0.982610</td>\n",
       "      <td>0.995430</td>\n",
       "      <td>...</td>\n",
       "      <td>0.028249</td>\n",
       "      <td>0.343503</td>\n",
       "      <td>0.120863</td>\n",
       "      <td>0.994389</td>\n",
       "      <td>0.997903</td>\n",
       "      <td>0.996143</td>\n",
       "      <td>0.746942</td>\n",
       "      <td>0.993803</td>\n",
       "      <td>0.997583</td>\n",
       "      <td>0.995689</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>stead_gpdpick_0.01</td>\n",
       "      <td>stead</td>\n",
       "      <td>gpdpick</td>\n",
       "      <td>0.0100</td>\n",
       "      <td>0</td>\n",
       "      <td>0.993886</td>\n",
       "      <td>0.990126</td>\n",
       "      <td>0.992002</td>\n",
       "      <td>0.994755</td>\n",
       "      <td>0.994433</td>\n",
       "      <td>...</td>\n",
       "      <td>0.036547</td>\n",
       "      <td>0.349581</td>\n",
       "      <td>0.122593</td>\n",
       "      <td>0.994505</td>\n",
       "      <td>0.997884</td>\n",
       "      <td>0.996191</td>\n",
       "      <td>0.759573</td>\n",
       "      <td>0.994015</td>\n",
       "      <td>0.997807</td>\n",
       "      <td>0.995907</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>stead_phasenet_0.0001</td>\n",
       "      <td>stead</td>\n",
       "      <td>phasenet</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>0</td>\n",
       "      <td>0.991224</td>\n",
       "      <td>0.993663</td>\n",
       "      <td>0.992442</td>\n",
       "      <td>0.960440</td>\n",
       "      <td>0.990832</td>\n",
       "      <td>...</td>\n",
       "      <td>0.008305</td>\n",
       "      <td>0.316787</td>\n",
       "      <td>0.121676</td>\n",
       "      <td>0.986430</td>\n",
       "      <td>0.993574</td>\n",
       "      <td>0.989989</td>\n",
       "      <td>0.495291</td>\n",
       "      <td>0.984863</td>\n",
       "      <td>0.993895</td>\n",
       "      <td>0.989358</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>stead_phasenet</td>\n",
       "      <td>stead</td>\n",
       "      <td>phasenet</td>\n",
       "      <td>0.0010</td>\n",
       "      <td>0</td>\n",
       "      <td>0.993606</td>\n",
       "      <td>0.994309</td>\n",
       "      <td>0.993958</td>\n",
       "      <td>0.993966</td>\n",
       "      <td>0.993651</td>\n",
       "      <td>...</td>\n",
       "      <td>0.008583</td>\n",
       "      <td>0.304581</td>\n",
       "      <td>0.118452</td>\n",
       "      <td>0.987781</td>\n",
       "      <td>0.995107</td>\n",
       "      <td>0.991431</td>\n",
       "      <td>0.470745</td>\n",
       "      <td>0.986771</td>\n",
       "      <td>0.995419</td>\n",
       "      <td>0.991076</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>stead_phasenet_0.01</td>\n",
       "      <td>stead</td>\n",
       "      <td>phasenet</td>\n",
       "      <td>0.0100</td>\n",
       "      <td>0</td>\n",
       "      <td>0.994034</td>\n",
       "      <td>0.994309</td>\n",
       "      <td>0.994172</td>\n",
       "      <td>1.055480</td>\n",
       "      <td>0.994395</td>\n",
       "      <td>...</td>\n",
       "      <td>0.008594</td>\n",
       "      <td>0.283973</td>\n",
       "      <td>0.111816</td>\n",
       "      <td>0.990847</td>\n",
       "      <td>0.994156</td>\n",
       "      <td>0.992499</td>\n",
       "      <td>0.488864</td>\n",
       "      <td>0.989855</td>\n",
       "      <td>0.994284</td>\n",
       "      <td>0.992064</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>104 rows × 31 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                   experiment   data         model      lr  version  \\\n",
       "0    ethz_basicphaseae_0.0001   ethz  basicphaseae  0.0001        0   \n",
       "1           ethz_basicphaseae   ethz  basicphaseae  0.0010        0   \n",
       "2      ethz_basicphaseae_0.01   ethz  basicphaseae  0.0100        0   \n",
       "3            ethz_cred_0.0001   ethz          cred  0.0001        0   \n",
       "4                   ethz_cred   ethz          cred  0.0010        0   \n",
       "..                        ...    ...           ...     ...      ...   \n",
       "99              stead_gpdpick  stead       gpdpick  0.0010        0   \n",
       "100        stead_gpdpick_0.01  stead       gpdpick  0.0100        0   \n",
       "101     stead_phasenet_0.0001  stead      phasenet  0.0001        0   \n",
       "102            stead_phasenet  stead      phasenet  0.0010        0   \n",
       "103       stead_phasenet_0.01  stead      phasenet  0.0100        0   \n",
       "\n",
       "     dev_phase_precision  dev_phase_recall  dev_phase_f1  phase_threshold  \\\n",
       "0               0.818765          0.891444      0.853560         7.936640   \n",
       "1               0.920499          0.967113      0.943230         2.620540   \n",
       "2                    NaN               NaN           NaN              NaN   \n",
       "3                    NaN               NaN           NaN              NaN   \n",
       "4                    NaN               NaN           NaN              NaN   \n",
       "..                   ...               ...           ...              ...   \n",
       "99              0.994568          0.991264      0.992913         0.982610   \n",
       "100             0.993886          0.990126      0.992002         0.994755   \n",
       "101             0.991224          0.993663      0.992442         0.960440   \n",
       "102             0.993606          0.994309      0.993958         0.993966   \n",
       "103             0.994034          0.994309      0.994172         1.055480   \n",
       "\n",
       "     test_phase_precision  ...  test_S_mean_s  test_S_std_s  test_S_mae_s  \\\n",
       "0                0.751454  ...       0.201656      1.309772      0.630289   \n",
       "1                0.868497  ...       0.065660      1.241713      0.533804   \n",
       "2                     NaN  ...            NaN           NaN           NaN   \n",
       "3                     NaN  ...            NaN           NaN           NaN   \n",
       "4                     NaN  ...            NaN           NaN           NaN   \n",
       "..                    ...  ...            ...           ...           ...   \n",
       "99               0.995430  ...       0.028249      0.343503      0.120863   \n",
       "100              0.994433  ...       0.036547      0.349581      0.122593   \n",
       "101              0.990832  ...       0.008305      0.316787      0.121676   \n",
       "102              0.993651  ...       0.008583      0.304581      0.118452   \n",
       "103              0.994395  ...       0.008594      0.283973      0.111816   \n",
       "\n",
       "     dev_det_precision  dev_det_recall  dev_det_f1  det_threshold  \\\n",
       "0             0.932951        0.762390    0.839091       0.316741   \n",
       "1             0.956396        0.911894    0.933615       0.296525   \n",
       "2             0.500000        1.000000    0.666667       0.000000   \n",
       "3             0.903446        0.945485    0.923988       0.077826   \n",
       "4             0.920923        0.923458    0.922189       0.061431   \n",
       "..                 ...             ...         ...            ...   \n",
       "99            0.994389        0.997903    0.996143       0.746942   \n",
       "100           0.994505        0.997884    0.996191       0.759573   \n",
       "101           0.986430        0.993574    0.989989       0.495291   \n",
       "102           0.987781        0.995107    0.991431       0.470745   \n",
       "103           0.990847        0.994156    0.992499       0.488864   \n",
       "\n",
       "     test_det_precision  test_det_recall  test_det_f1  \n",
       "0              0.921642         0.702241     0.797120  \n",
       "1              0.939023         0.885646     0.911554  \n",
       "2              0.000000         0.000000     0.000000  \n",
       "3              0.883105         0.935241     0.908426  \n",
       "4              0.890877         0.911779     0.901207  \n",
       "..                  ...              ...          ...  \n",
       "99             0.993803         0.997583     0.995689  \n",
       "100            0.994015         0.997807     0.995907  \n",
       "101            0.984863         0.993895     0.989358  \n",
       "102            0.986771         0.995419     0.991076  \n",
       "103            0.989855         0.994284     0.992064  \n",
       "\n",
       "[104 rows x 31 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "009f39b8-8bf6-4aed-b4c7-2834ec9b6604",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 instance gpd 0.0001\n",
      "0 instance gpdpick 0.0001\n",
      "0 neic gpd 0.0001\n",
      "0 neic gpdpick 0.0001\n"
     ]
    }
   ],
   "source": [
    "# Detect missing grid entries\n",
    "\n",
    "for data in results[\"data\"].unique():\n",
    "    for model in results[\"model\"].unique():\n",
    "        for lr in results[\"lr\"].unique():\n",
    "            mask = np.logical_and(results[\"data\"] == data, results[\"model\"] == model)\n",
    "            mask = np.logical_and(mask, results[\"lr\"] == lr)\n",
    "            if np.sum(mask) != 1:\n",
    "                print(np.sum(mask), data, model, lr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "vietnamese-sugar",
   "metadata": {},
   "outputs": [],
   "source": [
    "def results_to_array(results, cols, selection, minimize=False):\n",
    "    data_dict = {data: i for i, data in enumerate(results[\"data\"].unique())}\n",
    "    model_dict = {model: i for i, model in enumerate(results[\"model\"].unique())}\n",
    "\n",
    "    n_data = len(results[\"data\"].unique())\n",
    "    n_model = len(results[\"model\"].unique())\n",
    "    \n",
    "    res_array = np.nan * np.zeros((n_data, n_model, len(cols)))\n",
    "    \n",
    "    for _, subdf in results.groupby([\"data\", \"model\"]):\n",
    "        if np.isnan(subdf[selection]).all():\n",
    "            continue\n",
    "        if minimize:\n",
    "            i = np.nanargmin(subdf[selection])\n",
    "        else:\n",
    "            i = np.nanargmax(subdf[selection])\n",
    "        row = subdf.iloc[i]\n",
    "        entry = res_array[data_dict[row[\"data\"]], model_dict[row[\"model\"]]]\n",
    "        for col_idx, col in enumerate(cols):\n",
    "            entry[col_idx] = row[col]\n",
    "    \n",
    "    return data_dict, model_dict, res_array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "danish-discretion",
   "metadata": {},
   "outputs": [],
   "source": [
    "def results_to_table(results, cols, selection, labels=None, minimize=False, average=None):\n",
    "    data_dict, model_dict, res_array = results_to_array(results, cols, selection, minimize=minimize)\n",
    "    \n",
    "    if average is None:\n",
    "        average = list(range(len(cols)))\n",
    "    \n",
    "    n_data = len(results[\"data\"].unique())\n",
    "    n_model = len(results[\"model\"].unique())\n",
    "    \n",
    "    inv_data_dict = {v: k for k, v in data_dict.items()}\n",
    "    inv_model_dict = {v: k for k, v in model_dict.items()}\n",
    "\n",
    "    header = \"\"\n",
    "    header_count = 0\n",
    "    for j in range(n_model):\n",
    "        if np.isnan(res_array[:, j]).all():\n",
    "            print(f\"Skipping {inv_model_dict[j]}\")\n",
    "            continue\n",
    "        header += f\" & \\multicolumn{{{len(cols)}}}{{|c}}{{{inv_model_dict[j]}}}\"\n",
    "        header_count += 1\n",
    "    header += f\" & \\multicolumn{{{len(average)}}}{{|c}}{{average}} \\\\\\\\\"\n",
    "    \n",
    "    avg_labels = [label for i, label in enumerate(labels) if i in average]\n",
    "    \n",
    "    if labels is not None:\n",
    "        label_str = \" & \".join([\"\"] + header_count * labels + avg_labels) + \"\\\\\\\\\"\n",
    "    else:\n",
    "        label_str = \"\"\n",
    "    \n",
    "    colspec = \"|\".join([\"c\"] + header_count * [\"c\" * len(cols)] + [\"c\" * len(average)])\n",
    "    tabular = f\"\\\\begin{{tabular}}{{{colspec}}}\"\n",
    "    \n",
    "    table_str = [tabular, header, label_str, \"\\\\hline\"]\n",
    "    \n",
    "    for i in range(n_data):\n",
    "        line = inv_data_dict[i]\n",
    "        if np.isnan(res_array[i]).all():\n",
    "            print(f\"Skipping {inv_data_dict[i]}\")\n",
    "            continue\n",
    "        for j in range(n_model):\n",
    "            if np.isnan(res_array[:, j]).all():\n",
    "                print(f\"Skipping {inv_model_dict[j]}\")\n",
    "                continue\n",
    "            for col_idx, _ in enumerate(cols):\n",
    "                line += f\" & {res_array[i, j, col_idx]:.2f}\"\n",
    "        \n",
    "        avg = np.nanmean(res_array[i], axis=0)\n",
    "        for col_idx, _ in enumerate(cols):\n",
    "            if col_idx in average:\n",
    "                line += f\" & {avg[col_idx]:.2f}\"\n",
    "        \n",
    "        line += \" \\\\\\\\\"\n",
    "        table_str.append(line)\n",
    "    \n",
    "    table_str.append(\"\\\\hline\")\n",
    "    \n",
    "    line = \"average\"\n",
    "    for j in range(n_model):\n",
    "        if np.isnan(res_array[:, j]).all():\n",
    "            print(f\"Skipping {inv_model_dict[j]}\")\n",
    "            continue\n",
    "        avg = np.nanmean(res_array[:, j], axis=0)\n",
    "        \n",
    "        for col_idx, _ in enumerate(cols):\n",
    "            if col_idx in average:\n",
    "                line += f\" & {avg[col_idx]:.2f}\"\n",
    "            else:\n",
    "                line += \" &\"\n",
    "    \n",
    "    line += \" \\\\\\\\\"\n",
    "    table_str.append(line)    \n",
    "    table_str.append(\"\\\\end{tabular}\")\n",
    "    \n",
    "    return \"\\n\".join(table_str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "hungry-civilization",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Skipping neic\n",
      "\\begin{tabular}{c|ccc|ccc|ccc|ccc|ccc|ccc|ccc}\n",
      " & \\multicolumn{3}{|c}{basicphaseae} & \\multicolumn{3}{|c}{cred} & \\multicolumn{3}{|c}{eqtransformer} & \\multicolumn{3}{|c}{gpd} & \\multicolumn{3}{|c}{gpdpick} & \\multicolumn{3}{|c}{phasenet} & \\multicolumn{3}{|c}{average} \\\\\n",
      " & P & R & F1 & P & R & F1 & P & R & F1 & P & R & F1 & P & R & F1 & P & R & F1 & P & R & F1\\\\\n",
      "\\hline\n",
      "ethz & 0.94 & 0.89 & 0.91 & 0.88 & 0.94 & 0.91 & 0.88 & 0.94 & 0.91 & 0.93 & 0.98 & 0.95 & 0.95 & 0.98 & 0.96 & 0.93 & 0.96 & 0.95 & 0.92 & 0.95 & 0.93 \\\\\n",
      "geofon & 0.50 & 1.00 & 0.67 & 0.93 & 0.87 & 0.90 & 0.95 & 0.93 & 0.94 & 0.82 & 0.83 & 0.82 & 0.76 & 0.87 & 0.81 & 0.86 & 0.85 & 0.85 & 0.80 & 0.89 & 0.83 \\\\\n",
      "instance & 0.87 & 1.00 & 0.93 & 0.95 & 0.95 & 0.95 & 0.94 & 0.97 & 0.95 & 0.96 & 0.94 & 0.95 & 0.96 & 0.94 & 0.95 & 0.96 & 0.93 & 0.95 & 0.94 & 0.96 & 0.95 \\\\\n",
      "lendb & 0.53 & 0.98 & 0.68 & 0.98 & 0.94 & 0.96 & 0.98 & 0.93 & 0.96 & 0.93 & 0.84 & 0.88 & 0.91 & 0.86 & 0.89 & 0.93 & 0.88 & 0.90 & 0.88 & 0.90 & 0.88 \\\\\n",
      "stead & 0.98 & 0.99 & 0.98 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 0.99 & 1.00 & 1.00 & 0.99 & 0.99 & 0.99 & 0.99 & 1.00 & 0.99 \\\\\n",
      "\\hline\n",
      "average & 0.76 & 0.97 & 0.83 & 0.95 & 0.94 & 0.94 & 0.95 & 0.95 & 0.95 & 0.93 & 0.92 & 0.92 & 0.92 & 0.93 & 0.92 & 0.94 & 0.92 & 0.93 \\\\\n",
      "\\end{tabular}\n"
     ]
    }
   ],
   "source": [
    "table = results_to_table(results, [\"test_det_precision\", \"test_det_recall\", \"test_det_f1\"], \"dev_det_f1\", [\"P\", \"R\", \"F1\"])\n",
    "print(table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "amazing-judgment",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping lendb\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "\\begin{tabular}{c|ccc|ccc|ccc|ccc|ccc|ccc}\n",
      " & \\multicolumn{3}{|c}{basicphaseae} & \\multicolumn{3}{|c}{eqtransformer} & \\multicolumn{3}{|c}{gpd} & \\multicolumn{3}{|c}{gpdpick} & \\multicolumn{3}{|c}{phasenet} & \\multicolumn{3}{|c}{average} \\\\\n",
      " & P & R & F1 & P & R & F1 & P & R & F1 & P & R & F1 & P & R & F1 & P & R & F1\\\\\n",
      "\\hline\n",
      "ethz & 0.87 & 0.96 & 0.91 & 0.98 & 0.99 & 0.99 & 0.93 & 0.97 & 0.95 & 0.96 & 0.98 & 0.97 & 0.95 & 0.99 & 0.97 & 0.94 & 0.98 & 0.96 \\\\\n",
      "geofon & 0.99 & 1.00 & 0.99 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 1.00 & 0.99 & 1.00 & 0.99 & 0.99 & 0.99 & 0.99 & 1.00 & 0.99 \\\\\n",
      "instance & 0.91 & 0.97 & 0.94 & 0.96 & 0.98 & 0.97 & 0.92 & 0.97 & 0.95 & 0.95 & 0.98 & 0.97 & 0.94 & 0.98 & 0.96 & 0.94 & 0.98 & 0.96 \\\\\n",
      "neic & 0.91 & 0.97 & 0.94 & 0.99 & 0.99 & 0.99 & 0.95 & 0.98 & 0.96 & 0.95 & 0.98 & 0.96 & 0.94 & 0.98 & 0.96 & 0.94 & 0.98 & 0.96 \\\\\n",
      "stead & 0.94 & 0.97 & 0.96 & 1.00 & 1.00 & 1.00 & 0.97 & 0.97 & 0.97 & 1.00 & 0.99 & 0.99 & 0.99 & 0.99 & 0.99 & 0.98 & 0.99 & 0.98 \\\\\n",
      "\\hline\n",
      "average & 0.92 & 0.97 & 0.95 & 0.99 & 0.99 & 0.99 & 0.95 & 0.98 & 0.97 & 0.97 & 0.99 & 0.98 & 0.96 & 0.99 & 0.97 \\\\\n",
      "\\end{tabular}\n"
     ]
    }
   ],
   "source": [
    "table = results_to_table(results, [\"test_phase_precision\", \"test_phase_recall\", \"test_phase_f1\"], \"dev_phase_f1\", [\"P\", \"R\", \"F1\"])\n",
    "print(table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "consolidated-array",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping lendb\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "\\begin{tabular}{c|ccc|ccc|ccc|ccc|ccc|cc}\n",
      " & \\multicolumn{3}{|c}{basicphaseae} & \\multicolumn{3}{|c}{eqtransformer} & \\multicolumn{3}{|c}{gpd} & \\multicolumn{3}{|c}{gpdpick} & \\multicolumn{3}{|c}{phasenet} & \\multicolumn{2}{|c}{average} \\\\\n",
      " & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\sigma$ & MAE\\\\\n",
      "\\hline\n",
      "ethz & 0.11 & 0.77 & 0.30 & -0.02 & 0.35 & 0.11 & -0.22 & 0.55 & 0.38 & 0.05 & 0.42 & 0.12 & 0.02 & 0.42 & 0.12 & 0.50 & 0.21 \\\\\n",
      "geofon & 0.10 & 2.19 & 1.27 & 0.14 & 0.95 & 0.42 & 0.13 & 1.53 & 0.77 & 0.21 & 1.55 & 0.72 & 0.14 & 1.28 & 0.58 & 1.50 & 0.75 \\\\\n",
      "instance & -0.04 & 1.42 & 0.56 & -0.06 & 1.33 & 0.43 & -0.27 & 1.30 & 0.64 & -0.11 & 1.29 & 0.43 & -0.10 & 1.32 & 0.46 & 1.33 & 0.51 \\\\\n",
      "neic & 0.20 & 1.93 & 1.09 & 0.00 & 0.39 & 0.06 & 0.10 & 1.49 & 0.79 & 0.22 & 1.47 & 0.73 & 0.24 & 1.46 & 0.75 & 1.35 & 0.68 \\\\\n",
      "stead & 0.06 & 0.43 & 0.14 & -0.01 & 0.19 & 0.07 & -0.10 & 0.36 & 0.27 & 0.02 & 0.24 & 0.08 & 0.01 & 0.33 & 0.09 & 0.31 & 0.13 \\\\\n",
      "\\hline\n",
      "average & & 1.35 & 0.67 & & 0.64 & 0.22 & & 1.05 & 0.57 & & 0.99 & 0.42 & & 0.96 & 0.40 \\\\\n",
      "\\end{tabular}\n"
     ]
    }
   ],
   "source": [
    "table = results_to_table(results, [\"test_P_mean_s\", \"test_P_std_s\", \"test_P_mae_s\"], \"dev_P_std_s\", [\"$\\\\mu$\", \"$\\\\sigma$\", \"MAE\"], minimize=True, average=[1, 2])\n",
    "print(table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "terminal-recycling",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping lendb\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "Skipping cred\n",
      "\\begin{tabular}{c|ccc|ccc|ccc|ccc|ccc|cc}\n",
      " & \\multicolumn{3}{|c}{basicphaseae} & \\multicolumn{3}{|c}{eqtransformer} & \\multicolumn{3}{|c}{gpd} & \\multicolumn{3}{|c}{gpdpick} & \\multicolumn{3}{|c}{phasenet} & \\multicolumn{2}{|c}{average} \\\\\n",
      " & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\mu$ & $\\sigma$ & MAE & $\\sigma$ & MAE\\\\\n",
      "\\hline\n",
      "ethz & 0.07 & 1.24 & 0.53 & 0.06 & 0.49 & 0.15 & 0.37 & 0.79 & 0.54 & 0.13 & 0.57 & 0.20 & 0.10 & 0.56 & 0.19 & 0.73 & 0.32 \\\\\n",
      "geofon & 0.09 & 2.02 & 1.25 & 0.30 & 1.35 & 0.80 & -0.03 & 1.30 & 0.77 & 0.15 & 1.27 & 0.71 & 0.30 & 1.71 & 1.02 & 1.53 & 0.91 \\\\\n",
      "instance & 0.26 & 1.28 & 0.55 & 0.24 & 1.26 & 0.44 & 0.18 & 1.28 & 0.64 & 0.25 & 1.22 & 0.45 & 0.26 & 1.26 & 0.48 & 1.26 & 0.51 \\\\\n",
      "neic & 0.11 & 1.50 & 0.82 & 0.00 & 0.35 & 0.08 & 0.10 & 1.23 & 0.68 & 0.19 & 1.20 & 0.61 & 0.19 & 1.19 & 0.62 & 1.09 & 0.56 \\\\\n",
      "stead & 0.01 & 0.44 & 0.19 & -0.00 & 0.23 & 0.10 & -0.10 & 0.40 & 0.27 & 0.03 & 0.34 & 0.12 & 0.01 & 0.28 & 0.11 & 0.34 & 0.16 \\\\\n",
      "\\hline\n",
      "average & & 1.30 & 0.67 & & 0.73 & 0.31 & & 1.00 & 0.58 & & 0.92 & 0.42 & & 1.00 & 0.48 \\\\\n",
      "\\end{tabular}\n"
     ]
    }
   ],
   "source": [
    "table = results_to_table(results, [\"test_S_mean_s\", \"test_S_std_s\", \"test_S_mae_s\"], \"dev_S_std_s\", [\"$\\\\mu$\", \"$\\\\sigma$\", \"MAE\"], minimize=True, average=[1, 2])\n",
    "print(table)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "arctic-vault",
   "metadata": {},
   "source": [
    "## Residual plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "synthetic-disaster",
   "metadata": {},
   "outputs": [],
   "source": [
    "def residual_matrix(phase, results, pred_path, selection, lim=2):\n",
    "    data_dict, model_dict, res_array = results_to_array(results, [f\"test_{phase}_mean_s\"], selection, minimize=True)\n",
    "    res_array = res_array[:, :, 0]\n",
    "    \n",
    "    n_data = len(results[\"data\"].unique())\n",
    "    n_model = len(results[\"model\"].unique())\n",
    "    \n",
    "    inv_data_dict = {v: k for k, v in data_dict.items()}\n",
    "    inv_model_dict = {v: k for k, v in model_dict.items()}\n",
    "    \n",
    "    true_n_data = np.sum((~np.isnan(res_array)).any(axis=1))\n",
    "    true_n_model = np.sum((~np.isnan(res_array)).any(axis=0))\n",
    "    \n",
    "    fig = plt.figure(figsize=(15, 15))\n",
    "    axs = fig.subplots(true_n_data, true_n_model, sharex=True, sharey=False, gridspec_kw={\"hspace\": 0.05, \"wspace\": 0.05})\n",
    "    \n",
    "    true_i = 0\n",
    "    for i in range(n_data):\n",
    "        if np.isnan(res_array[i]).all():\n",
    "            continue\n",
    "        true_j = 0\n",
    "        for j in range(n_model):\n",
    "            if np.isnan(res_array[:, j]).all():\n",
    "                continue\n",
    "            data, model = inv_data_dict[i], inv_model_dict[j]\n",
    "            mask = np.logical_and(results[\"model\"] == model, results[\"data\"] == data)\n",
    "            \n",
    "            subdf = results[mask]\n",
    "            lr_idx = np.nanargmin(subdf[selection])\n",
    "            row = subdf.iloc[lr_idx]\n",
    "            \n",
    "            axs[true_i, true_j].set_yticklabels([])\n",
    "            axs[true_i, true_j].set_xlim(-lim, lim)\n",
    "            \n",
    "            pred_path_loc = pred_path / row[\"experiment\"] / f\"version_{row['version']}\" / \"test_task23.csv\"\n",
    "            if not pred_path_loc.is_file():\n",
    "                true_j += 1\n",
    "                continue\n",
    "            pred = pd.read_csv(pred_path_loc)\n",
    "            \n",
    "            pred_phase = pred[pred[\"phase_label\"] == phase]\n",
    "            pred_col = f\"{phase.lower()}_sample_pred\"\n",
    "\n",
    "            diff = (pred_phase[pred_col] - pred_phase[\"phase_onset\"]) / pred_phase[\"sampling_rate\"]\n",
    "            \n",
    "            bins = np.linspace(-lim, lim, 50)\n",
    "            axs[true_i, true_j].hist(diff, bins=bins)\n",
    "            \n",
    "            axs[0, true_j].set_title(model)\n",
    "            \n",
    "            outliers = np.abs(diff) > lim\n",
    "            frac_outliers = np.sum(outliers) / len(diff)\n",
    "            diff[outliers] = lim\n",
    "            std_outlierfree = np.std(diff)\n",
    "            axs[true_i, true_j].text(0.98, 0.98, f\"{frac_outliers:.2f}\", transform=axs[true_i, true_j].transAxes, ha=\"right\", va=\"top\")\n",
    "            # axs[true_i, true_j].text(0.98, 0.87, f\"{std_outlierfree:.2f}\", transform=axs[true_i, true_j].transAxes, ha=\"right\", va=\"top\")\n",
    "            \n",
    "            true_j += 1\n",
    "            \n",
    "        axs[true_i, 0].set_ylabel(data)\n",
    "        \n",
    "        true_i += 1\n",
    "    \n",
    "    for ax in axs[-1]:\n",
    "        ax.set_xlabel(\"$t_{pred} - t_{true}$\")\n",
    "    \n",
    "    return fig"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "perceived-alliance",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 25 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = residual_matrix(\"P\", results, Path(\"pred\"), \"dev_P_std_s\")\n",
    "fig.savefig(\"test_P_diff.eps\", bbox_inches=\"tight\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "numeric-europe",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 25 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = residual_matrix(\"S\", results, Path(\"pred\"), \"dev_S_std_s\")\n",
    "fig.savefig(\"test_S_diff.eps\", bbox_inches=\"tight\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b0f7dfd2-cff1-44b4-bc6b-1fd2cde52d98",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}