simple_regression_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "bc4e060a-abbb-4ac3-afde-d88694183b56",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-08T23:19:38.807811Z",
     "iopub.status.busy": "2022-08-08T23:19:38.806879Z",
     "iopub.status.idle": "2022-08-08T23:19:43.012320Z",
     "shell.execute_reply": "2022-08-08T23:19:43.011901Z",
     "shell.execute_reply.started": "2022-08-08T23:19:38.807727Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import glob\n",
    "import os\n",
    "import seaborn as sns\n",
    "from tqdm import tqdm\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.layers import Dense\n",
    "from keras.models import Sequential\n",
    "from keras.wrappers.scikit_learn import KerasRegressor\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b23b1d8d-d270-46e6-8dae-1b56771b23fa",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true,
    "tags": []
   },
   "source": [
    "# Reload a trained model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1860b4f1-6626-4e3c-b967-fab84a2cbd68",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "### Load data\n",
    "train = pd.read_hdf(\"train_dnn.h5\")\n",
    "val = pd.read_hdf(\"val_dnn.h5\")\n",
    "test = pd.read_hdf(\"test_dnn.h5\")\n",
    "test.reset_index(inplace=True)\n",
    "\n",
    "### All clusters + track eta/phi/z0\n",
    "train_vars = [var for var in test.keys() if var.startswith('log10') and var != 'log10_truth_particle_e']\n",
    "train_vars += ['track_eta', 'track_phi', 'track_z0']\n",
    "\n",
    "train_x = train[train_vars].values\n",
    "train_y = train['log10_truth_particle_e'].values\n",
    "val_x = val[train_vars].values\n",
    "val_y = val['log10_truth_particle_e'].values\n",
    "test_x = test[train_vars].values\n",
    "test_y = test['log10_truth_particle_e'].values\n",
    "\n",
    "### Normalize the inputs \n",
    "sc = StandardScaler()\n",
    "train_x = sc.fit_transform(train_x)\n",
    "val_x = sc.transform(val_x)\n",
    "test_x = sc.transform(test_x)\n",
    "\n",
    "def regression_model():\n",
    "    model = Sequential()\n",
    "    model.add(Dense(50, input_dim=train_x.shape[1], activation='relu'))\n",
    "    model.add(Dense(50, activation='relu'))\n",
    "    model.add(Dense(50, activation='relu'))\n",
    "    model.add(Dense(1))\n",
    "    model.compile(loss='mean_absolute_error', optimizer='adam')\n",
    "    return model\n",
    "\n",
    "model = regression_model()\n",
    "model.load_weights(\"dnn_best_weights_50.h5\")\n",
    "test['nn_output'] = 10**model.predict(test_x) # re-exponentiate the energy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ddc7bde2-d1a6-400f-bf3a-480ce7346bff",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "### Response median plot \n",
    "import scipy.stats as stats\n",
    "from matplotlib.colors import ListedColormap\n",
    "from matplotlib.colors import LogNorm\n",
    "\n",
    "x = test.truth_particle_e\n",
    "y = (test.nn_output)/x\n",
    "\n",
    "xbin = [10**exp for exp in np.arange(-1., 3.1, 0.1)]\n",
    "ybin = np.arange(0., 3.1, 0.05)\n",
    "xcenter = [(xbin[i] + xbin[i+1]) / 2 for i in range(len(xbin)-1)]\n",
    "profileXMed = stats.binned_statistic(\n",
    "    x, y, bins=xbin, statistic='median').statistic\n",
    "\n",
    "c_map = ListedColormap(sns.color_palette(\"Blues\", n_colors=100).as_hex())\n",
    "fig = plt.figure(figsize=(12,8), dpi=200)\n",
    "fig.patch.set_facecolor('white')\n",
    "plt.hist2d(x, y, bins=[xbin, ybin], norm=LogNorm(),zorder = -1, cmap=c_map);\n",
    "plt.plot(np.array(xcenter), np.array(profileXMed), color='indianred', linewidth=3)\n",
    "plt.plot([0.1, 1000], [1, 1], linestyle='--', color='black');\n",
    "plt.xscale('log')\n",
    "plt.ylim(0, 1.75)\n",
    "plt.xlim(0.3, )\n",
    "plt.xlabel('Truth Particle Energy [GeV]')\n",
    "plt.ylabel('Predicted Energy / Target');\n",
    "np.savez('pub_note_results/response_median_dnn_50.npz', response_median=profileXMed, xcenter=xcenter)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0e5e9dd2-043e-4645-aff2-e95e5d7c6c76",
   "metadata": {},
   "outputs": [],
   "source": [
    "### IQR plot \n",
    "\n",
    "def iqrOverMed(x):\n",
    "    # get the IQR via the percentile function\n",
    "    # 84 is median + 1 sigma, 16 is median - 1 sigma\n",
    "    q84, q16 = np.percentile(x, [84, 16])\n",
    "    iqr = q84 - q16\n",
    "    med = np.median(x)\n",
    "    return iqr / (2*med)\n",
    "\n",
    "xbin = [10**exp for exp in np.arange(-1., 3.1, 0.1)]\n",
    "xcenter = [(xbin[i] + xbin[i+1]) / 2 for i in range(len(xbin)-1)]\n",
    "resolution = stats.binned_statistic(x, y, bins=xbin, statistic=iqrOverMed).statistic\n",
    "\n",
    "fig = plt.figure(figsize=(10,6), dpi=200)\n",
    "fig.patch.set_facecolor('white')\n",
    "plt.plot(xcenter, resolution, linewidth=3)\n",
    "plt.xscale('log')\n",
    "plt.xlim(0.1, 1000)\n",
    "plt.ylim(0,0.5)\n",
    "plt.xlabel('Truth Particle Energy [GeV]')\n",
    "plt.ylabel('Response IQR / 2 x Median');\n",
    "\n",
    "np.savez('pub_note_results/iqr_dnn_50.npz', iqr=resolution, xcenter=xcenter)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a8710069-82f1-4d72-ae67-308c59fa2d8d",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "np.min(test.nn_output/test.truth_particle_e)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8118c192-23ac-4c0f-be4c-69ce63ae12b4",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "test[(test.truth_particle_e > 100) & (test.truth_particle_e < 101) & \n",
    "   (test.nn_output/test.truth_particle_e < 0.2)][[\"nn_output\", \"truth_particle_e\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "676d9d9a-b7e4-4674-91a8-16d68d3ea9f6",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "plt.hist(test.nn_output/test.truth_particle_e, bins=np.linspace(0,2,100));"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b3da379-cb13-4b52-90e8-ee24866bd6c8",
   "metadata": {},
   "source": [
    "Now compare to Transformer event..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d3e50b5d-347f-492f-8a68-8add116f0e31",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "transformer = np.load('pub_note_results/Attention_Prediction_TRK_MarielFile.npz') # as of July 24\n",
    "df_transformer = pd.DataFrame(np.vstack([transformer['target'],transformer['prediction']]).T, columns=[\"targets_regress\", \"outputs_regress\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9a15d650-7f18-4d34-a829-de4444cad0e0",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "len(test) == len(df_transformer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5b5c7691-3dbb-498f-8082-1dbd06098b18",
   "metadata": {},
   "outputs": [],
   "source": [
    "test.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fe4636dd-c39f-4ca2-83a1-d2854543ab16",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "test.truth_particle_e"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f28f1f00-56eb-4141-9ef0-73a6a4989497",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_transformer.targets_regress"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "37750c9f-dde0-432c-87cc-a154d1a4f90a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "np.sum(test.truth_particle_e == df_transformer.targets_regress)/len(test) \n",
    "# okay, so these are definitely identical events!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1831778a-ba08-419d-b2af-1c92111c4a94",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "np.min(df_transformer.outputs_regress/df_transformer.targets_regress)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "88557254-7dd9-460b-a23b-78b035876bb7",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.hist(df_transformer.outputs_regress/df_transformer.targets_regress, bins=np.linspace(0,2,100));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "89c5e90e-011f-4d7f-adb2-e3af5b4881fb",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "test[(np.abs(test.truth_particle_e - 1000) < 5)  # close to 100 GeV\n",
    "     & \n",
    "     (test.nn_output/test.truth_particle_e < 0.5)  # DNN does badly\n",
    "    ][[\"nn_output\", \"truth_particle_e\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fd556b47-b88c-49f7-a27f-c4ad317a9cdb",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "test.iloc[970158].nn_output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "507973c1-4515-4ec4-b8e2-cd936bf27049",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_transformer.iloc[970158]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "922e409c-1d4d-40a9-b334-8c39c9017f17",
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = [73920, 379097, 585445, 966165, 278235, 970158]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5d3b8b34-084f-467e-9aca-54e62ff3c048",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "test.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3e14a855-2e51-44f4-96bc-bfc41ffbe183",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "test.iloc[indices].truth_particle_e"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f5a8553d-1587-4cd5-bd22-0c16a6c055ce",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "test.iloc[indices][[\"truth_particle_e\",\"track_pt\", \"track_eta\", \"track_phi\", \"track_z0\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "80b522f7-bb8e-4d9f-9904-5ea7bb3b175a",
   "metadata": {},
   "outputs": [],
   "source": [
    "test.iloc[indices][[\"truth_particle_e\",\"cluster_e_0\", \"cluster_e_1\", \"cluster_e_2\", \"cluster_e_3\"]]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74472587-b9b8-4283-8fb6-e81e7f850adf",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Load data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ddef1fc4-b6b8-4320-b97c-b1bf3e432338",
   "metadata": {},
   "source": [
    "### Premade"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "eb50bd9d-2b4e-4e7f-aaa5-33e6cba63776",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_train = pd.read_hdf(\"train_dnn.h5\")\n",
    "df_val = pd.read_hdf(\"val_dnn.h5\")\n",
    "df_test = pd.read_hdf(\"test_dnn.h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "47e34dfb-3f2d-4d7c-afb6-470fe47d3fac",
   "metadata": {
    "tags": []
   },
   "source": [
    "### or, remake from scratch:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "fec78b14-f57d-4074-988d-46ec6f3fae87",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-08T23:22:57.370622Z",
     "iopub.status.busy": "2022-08-08T23:22:57.370442Z",
     "iopub.status.idle": "2022-08-08T23:24:14.618675Z",
     "shell.execute_reply": "2022-08-08T23:24:14.617896Z",
     "shell.execute_reply.started": "2022-08-08T23:22:57.370604Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 30/30 [00:45<00:00,  1.52s/it]\n",
      "100%|██████████| 10/10 [00:15<00:00,  1.57s/it]\n",
      "100%|██████████| 10/10 [00:14<00:00,  1.48s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training dataframe has 308,717 events.\n",
      "Validation dataframe has 102,715 events.\n",
      "Testing dataframe has 102,803 events.\n"
     ]
    }
   ],
   "source": [
    "files = glob.glob(\"../data/onetrack_multicluster_sanmay/pion_files/*.npy\")\n",
    "train = pd.concat([pd.DataFrame(np.load(file, allow_pickle=True).item()) for file in tqdm(files[:30])])\n",
    "val = pd.concat([pd.DataFrame(np.load(file, allow_pickle=True).item()) for file in tqdm(files[30:40])])\n",
    "test = pd.concat([pd.DataFrame(np.load(file, allow_pickle=True).item()) for file in tqdm(files[40:50])])\n",
    "\n",
    "# train = pd.concat([pd.DataFrame(np.load(file, allow_pickle=True).item()) for file in tqdm(files[:350])])\n",
    "# val = pd.concat([pd.DataFrame(np.load(file, allow_pickle=True).item()) for file in tqdm(files[350:400])])\n",
    "# test = pd.concat([pd.DataFrame(np.load(file, allow_pickle=True).item()) for file in tqdm(files[400:])])\n",
    "print(\"Training dataframe has {:,} events.\".format(train.shape[0]))\n",
    "print(\"Validation dataframe has {:,} events.\".format(val.shape[0]))\n",
    "print(\"Testing dataframe has {:,} events.\".format(test.shape[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "05a043f4-ec60-491a-adb1-26b647411b44",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-08T23:45:04.775554Z",
     "iopub.status.busy": "2022-08-08T23:45:04.775201Z",
     "iopub.status.idle": "2022-08-08T23:45:04.781861Z",
     "shell.execute_reply": "2022-08-08T23:45:04.781052Z",
     "shell.execute_reply.started": "2022-08-08T23:45:04.775517Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "### SANMAY'S VARIABLES\n",
    "cluster_var = ['cluster_EM_PROBABILITY', 'cluster_HAD_WEIGHT', 'cluster_OOC_WEIGHT',\n",
    "               'cluster_DM_WEIGHT', 'cluster_CENTER_MAG', 'cluster_FIRST_ENG_DENS', \n",
    "               'cluster_CENTER_LAMBDA', 'cluster_ISOLATION'\n",
    "              ]\n",
    "\n",
    "track_var = ['trackPt',\n",
    "             'trackP',\n",
    "             'trackMass',\n",
    "             'trackEta',\n",
    "             'trackPhi',\n",
    "             'trackNumberOfPixelHits',\n",
    "             'trackNumberOfSCTHits',\n",
    "             'trackNumberOfPixelDeadSensors',\n",
    "             'trackNumberOfSCTDeadSensors',\n",
    "             'trackNumberOfInnermostPixelLayerHits',\n",
    "             'trackNumberOfNextToInnermostPixelLayerHits',\n",
    "             'trackExpectInnermostPixelLayerHit',\n",
    "             'trackExpectNextToInnermostPixelLayerHit',\n",
    "             'trackNumberOfTRTHits',\n",
    "             'trackNumberOfTRTOutliers',\n",
    "             'trackChiSquared',\n",
    "             'trackNumberDOF',\n",
    "             'trackD0',\n",
    "             'trackZ0'\n",
    "            ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "fa64d529-d47c-41b4-88a1-aa6d3c35deb9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T00:05:45.682294Z",
     "iopub.status.busy": "2022-08-09T00:05:45.681486Z",
     "iopub.status.idle": "2022-08-09T00:05:46.681633Z",
     "shell.execute_reply": "2022-08-09T00:05:46.681073Z",
     "shell.execute_reply.started": "2022-08-09T00:05:45.682254Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "df1 = pd.DataFrame(pd.DataFrame(train.cluster_EM_PROBABILITY.to_list(), columns=[\"cluster_em_prob_\"+str(x) for x in np.arange(33)]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "b17f0f2e-0e6c-49dd-9d5d-470811b0832f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T00:06:03.007244Z",
     "iopub.status.busy": "2022-08-09T00:06:03.006938Z",
     "iopub.status.idle": "2022-08-09T00:06:03.969776Z",
     "shell.execute_reply": "2022-08-09T00:06:03.969048Z",
     "shell.execute_reply.started": "2022-08-09T00:06:03.007215Z"
    }
   },
   "outputs": [],
   "source": [
    "df2 = pd.DataFrame(pd.DataFrame(train.cluster_CENTER_LAMBDA.to_list(), columns=[\"cluster_CENTER_LAMBDA_\"+str(x) for x in np.arange(33)]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "3f5747c6-bcf9-426c-a268-3b9ecb3513d8",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T00:08:18.614288Z",
     "iopub.status.busy": "2022-08-09T00:08:18.613723Z",
     "iopub.status.idle": "2022-08-09T00:08:19.031516Z",
     "shell.execute_reply": "2022-08-09T00:08:19.030967Z",
     "shell.execute_reply.started": "2022-08-09T00:08:18.614257Z"
    }
   },
   "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>cluster_em_prob_0</th>\n",
       "      <th>cluster_em_prob_1</th>\n",
       "      <th>cluster_em_prob_2</th>\n",
       "      <th>cluster_em_prob_3</th>\n",
       "      <th>cluster_em_prob_4</th>\n",
       "      <th>cluster_em_prob_5</th>\n",
       "      <th>cluster_em_prob_6</th>\n",
       "      <th>cluster_em_prob_7</th>\n",
       "      <th>cluster_em_prob_8</th>\n",
       "      <th>cluster_em_prob_9</th>\n",
       "      <th>...</th>\n",
       "      <th>cluster_CENTER_LAMBDA_23</th>\n",
       "      <th>cluster_CENTER_LAMBDA_24</th>\n",
       "      <th>cluster_CENTER_LAMBDA_25</th>\n",
       "      <th>cluster_CENTER_LAMBDA_26</th>\n",
       "      <th>cluster_CENTER_LAMBDA_27</th>\n",
       "      <th>cluster_CENTER_LAMBDA_28</th>\n",
       "      <th>cluster_CENTER_LAMBDA_29</th>\n",
       "      <th>cluster_CENTER_LAMBDA_30</th>\n",
       "      <th>cluster_CENTER_LAMBDA_31</th>\n",
       "      <th>cluster_CENTER_LAMBDA_32</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.001000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.208537</td>\n",
       "      <td>0.000131</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.002079</td>\n",
       "      <td>0.000447</td>\n",
       "      <td>0.006793</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.000479</td>\n",
       "      <td>0.000402</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.002325</td>\n",
       "      <td>0.000953</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>308712</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>308713</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>308714</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>308715</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>308716</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>617434 rows × 66 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        cluster_em_prob_0  cluster_em_prob_1  cluster_em_prob_2  \\\n",
       "0                0.001000                NaN                NaN   \n",
       "1                0.208537           0.000131                NaN   \n",
       "2                0.002079           0.000447           0.006793   \n",
       "3                0.000479           0.000402                NaN   \n",
       "4                0.002325           0.000953                NaN   \n",
       "...                   ...                ...                ...   \n",
       "308712                NaN                NaN                NaN   \n",
       "308713                NaN                NaN                NaN   \n",
       "308714                NaN                NaN                NaN   \n",
       "308715                NaN                NaN                NaN   \n",
       "308716                NaN                NaN                NaN   \n",
       "\n",
       "        cluster_em_prob_3  cluster_em_prob_4  cluster_em_prob_5  \\\n",
       "0                     NaN                NaN                NaN   \n",
       "1                     NaN                NaN                NaN   \n",
       "2                     NaN                NaN                NaN   \n",
       "3                     NaN                NaN                NaN   \n",
       "4                     NaN                NaN                NaN   \n",
       "...                   ...                ...                ...   \n",
       "308712                NaN                NaN                NaN   \n",
       "308713                NaN                NaN                NaN   \n",
       "308714                NaN                NaN                NaN   \n",
       "308715                NaN                NaN                NaN   \n",
       "308716                NaN                NaN                NaN   \n",
       "\n",
       "        cluster_em_prob_6  cluster_em_prob_7  cluster_em_prob_8  \\\n",
       "0                     NaN                NaN                NaN   \n",
       "1                     NaN                NaN                NaN   \n",
       "2                     NaN                NaN                NaN   \n",
       "3                     NaN                NaN                NaN   \n",
       "4                     NaN                NaN                NaN   \n",
       "...                   ...                ...                ...   \n",
       "308712                NaN                NaN                NaN   \n",
       "308713                NaN                NaN                NaN   \n",
       "308714                NaN                NaN                NaN   \n",
       "308715                NaN                NaN                NaN   \n",
       "308716                NaN                NaN                NaN   \n",
       "\n",
       "        cluster_em_prob_9  ...  cluster_CENTER_LAMBDA_23  \\\n",
       "0                     NaN  ...                       NaN   \n",
       "1                     NaN  ...                       NaN   \n",
       "2                     NaN  ...                       NaN   \n",
       "3                     NaN  ...                       NaN   \n",
       "4                     NaN  ...                       NaN   \n",
       "...                   ...  ...                       ...   \n",
       "308712                NaN  ...                       NaN   \n",
       "308713                NaN  ...                       NaN   \n",
       "308714                NaN  ...                       NaN   \n",
       "308715                NaN  ...                       NaN   \n",
       "308716                NaN  ...                       NaN   \n",
       "\n",
       "        cluster_CENTER_LAMBDA_24  cluster_CENTER_LAMBDA_25  \\\n",
       "0                            NaN                       NaN   \n",
       "1                            NaN                       NaN   \n",
       "2                            NaN                       NaN   \n",
       "3                            NaN                       NaN   \n",
       "4                            NaN                       NaN   \n",
       "...                          ...                       ...   \n",
       "308712                       NaN                       NaN   \n",
       "308713                       NaN                       NaN   \n",
       "308714                       NaN                       NaN   \n",
       "308715                       NaN                       NaN   \n",
       "308716                       NaN                       NaN   \n",
       "\n",
       "        cluster_CENTER_LAMBDA_26  cluster_CENTER_LAMBDA_27  \\\n",
       "0                            NaN                       NaN   \n",
       "1                            NaN                       NaN   \n",
       "2                            NaN                       NaN   \n",
       "3                            NaN                       NaN   \n",
       "4                            NaN                       NaN   \n",
       "...                          ...                       ...   \n",
       "308712                       NaN                       NaN   \n",
       "308713                       NaN                       NaN   \n",
       "308714                       NaN                       NaN   \n",
       "308715                       NaN                       NaN   \n",
       "308716                       NaN                       NaN   \n",
       "\n",
       "        cluster_CENTER_LAMBDA_28  cluster_CENTER_LAMBDA_29  \\\n",
       "0                            NaN                       NaN   \n",
       "1                            NaN                       NaN   \n",
       "2                            NaN                       NaN   \n",
       "3                            NaN                       NaN   \n",
       "4                            NaN                       NaN   \n",
       "...                          ...                       ...   \n",
       "308712                       NaN                       NaN   \n",
       "308713                       NaN                       NaN   \n",
       "308714                       NaN                       NaN   \n",
       "308715                       NaN                       NaN   \n",
       "308716                       NaN                       NaN   \n",
       "\n",
       "        cluster_CENTER_LAMBDA_30  cluster_CENTER_LAMBDA_31  \\\n",
       "0                            NaN                       NaN   \n",
       "1                            NaN                       NaN   \n",
       "2                            NaN                       NaN   \n",
       "3                            NaN                       NaN   \n",
       "4                            NaN                       NaN   \n",
       "...                          ...                       ...   \n",
       "308712                       NaN                       NaN   \n",
       "308713                       NaN                       NaN   \n",
       "308714                       NaN                       NaN   \n",
       "308715                       NaN                       NaN   \n",
       "308716                       NaN                       NaN   \n",
       "\n",
       "        cluster_CENTER_LAMBDA_32  \n",
       "0                            NaN  \n",
       "1                            NaN  \n",
       "2                            NaN  \n",
       "3                            NaN  \n",
       "4                            NaN  \n",
       "...                          ...  \n",
       "308712                       NaN  \n",
       "308713                       NaN  \n",
       "308714                       NaN  \n",
       "308715                       NaN  \n",
       "308716                       NaN  \n",
       "\n",
       "[617434 rows x 66 columns]"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.concat([df1, df2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "50d665a7-4ecf-4b72-b95e-f9b2cb580816",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T00:12:30.701274Z",
     "iopub.status.busy": "2022-08-09T00:12:30.700805Z",
     "iopub.status.idle": "2022-08-09T00:12:30.717614Z",
     "shell.execute_reply": "2022-08-09T00:12:30.716536Z",
     "shell.execute_reply.started": "2022-08-09T00:12:30.701240Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def process(df): \n",
    "    ### Start the dataframe of inputs \n",
    "    ncols = pd.DataFrame(pd.DataFrame(df.cluster_E.to_list())).shape[1]\n",
    "    df2 = pd.DataFrame(pd.DataFrame(df.cluster_E.to_list(), columns=[\"cluster_e_\"+str(x) for x in np.arange(ncols)]))\n",
    "    \n",
    "    ### Add Sanmay's cluster inputs\n",
    "    dfs = []\n",
    "    for var in cluster_var:\n",
    "        dfs.append(pd.DataFrame(pd.DataFrame(df[var].to_list(), columns=[str(var)+str(x) for x in np.arange(ncols)])))\n",
    "    df_cluster = pd.concat(dfs, axis=1)\n",
    "    df2 = pd.concat([df2, df_cluster], axis=1)\n",
    "    \n",
    "    ### Add Sanmay's track inputs \n",
    "    for var in track_var:\n",
    "        df2[var] = np.array(df[var].explode())\n",
    "    \n",
    "    ### Add track pT & truth particle E \n",
    "    track_pt = np.array(df.trackPt.explode())\n",
    "    truth_particle_e = np.array(df.truthPartE.explode())\n",
    "    track_eta = np.array(df.trackEta.explode())\n",
    "    track_phi = np.array(df.trackPhi.explode())\n",
    "    track_z0 = np.array(df.trackZ0.explode())\n",
    "\n",
    "    df2[\"track_pt\"] = track_pt\n",
    "    df2[\"track_eta\"] = track_eta\n",
    "    df2[\"track_phi\"] = track_phi\n",
    "    df2[\"track_z0\"] = track_z0\n",
    "    df2[\"truth_particle_e\"] = truth_particle_e\n",
    "\n",
    "    ### Drop infs/NaNs \n",
    "    df2.replace([np.inf, -np.inf], np.nan, inplace=True)\n",
    "    df2 = df2.fillna(0)\n",
    "\n",
    "    ### Cluster_E > 0.5\n",
    "    df2 = df2[df2.cluster_e_0 > 0.5]\n",
    "\n",
    "    ### Lose outliers in track pT \n",
    "    df2 = df2[df2.track_pt < 5000]\n",
    "\n",
    "    ### Cast as float\n",
    "    df2 = df2.astype('float32')\n",
    "\n",
    "    ### Add the log of all energy variables\n",
    "    for var in df2.keys():\n",
    "        if var == \"track_pt\" or var == \"truth_particle_e\" or var.startswith(\"cluster_e_\"):\n",
    "            df2['log10_'+var] = np.log10(df2[var])\n",
    "        else:\n",
    "            continue\n",
    "\n",
    "    ### Do this again? \n",
    "    df2.replace([np.inf, -np.inf], np.nan, inplace=True)\n",
    "    df2 = df2.fillna(0)\n",
    "    return df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "d35c750e-8d7f-414a-a6b8-9e5475b2bd66",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T00:12:31.249741Z",
     "iopub.status.busy": "2022-08-09T00:12:31.248961Z",
     "iopub.status.idle": "2022-08-09T00:13:48.365505Z",
     "shell.execute_reply": "2022-08-09T00:13:48.364985Z",
     "shell.execute_reply.started": "2022-08-09T00:12:31.249707Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/321 [00:00<?, ?it/s]/clusterfs/ml4hep/mpettee/miniconda3/envs/nbdev/lib/python3.9/site-packages/pandas/core/arraylike.py:364: RuntimeWarning: divide by zero encountered in log10\n",
      "  result = getattr(ufunc, method)(*inputs, **kwargs)\n",
      "100%|██████████| 321/321 [00:00<00:00, 3062.29it/s]\n",
      "100%|██████████| 285/285 [00:00<00:00, 7444.03it/s]\n",
      "100%|██████████| 294/294 [00:00<00:00, 7463.85it/s]\n"
     ]
    }
   ],
   "source": [
    "df_train = process(train)\n",
    "df_val = process(val)\n",
    "df_test = process(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "id": "d8d808fb-3c30-4f4f-b0f0-bf05fe80f208",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T04:18:24.152314Z",
     "iopub.status.busy": "2022-08-09T04:18:24.152003Z",
     "iopub.status.idle": "2022-08-09T04:18:27.934920Z",
     "shell.execute_reply": "2022-08-09T04:18:27.934033Z",
     "shell.execute_reply.started": "2022-08-09T04:18:24.152284Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_train.to_hdf(\"train_dnn_sanmay.h5\", \"df\")\n",
    "df_val.to_hdf(\"val_dnn_sanmay.h5\", \"df\")\n",
    "df_test.to_hdf(\"test_dnn_sanmay.h5\", \"df\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "f2efe5dc-52c7-4032-a7d0-14949b200b69",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T00:15:14.760488Z",
     "iopub.status.busy": "2022-08-09T00:15:14.759789Z",
     "iopub.status.idle": "2022-08-09T00:15:14.768245Z",
     "shell.execute_reply": "2022-08-09T00:15:14.767514Z",
     "shell.execute_reply.started": "2022-08-09T00:15:14.760449Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['cluster_e_0', 'cluster_e_1', 'cluster_e_2', 'cluster_e_3',\n",
       "       'cluster_e_4', 'cluster_e_5', 'cluster_e_6', 'cluster_e_7',\n",
       "       'cluster_e_8', 'cluster_e_9', 'cluster_e_10', 'cluster_e_11',\n",
       "       'cluster_e_12', 'cluster_e_13', 'cluster_e_14', 'cluster_e_15',\n",
       "       'cluster_e_16', 'cluster_e_17', 'cluster_e_18', 'cluster_e_19',\n",
       "       'cluster_e_20', 'cluster_e_21', 'cluster_e_22', 'cluster_e_23',\n",
       "       'cluster_e_24', 'cluster_e_25', 'cluster_e_26', 'cluster_e_27',\n",
       "       'cluster_e_28', 'cluster_e_29', 'cluster_e_30', 'cluster_e_31',\n",
       "       'cluster_e_32', 'cluster_EM_PROBABILITY0',\n",
       "       'cluster_EM_PROBABILITY1', 'cluster_EM_PROBABILITY2',\n",
       "       'cluster_EM_PROBABILITY3', 'cluster_EM_PROBABILITY4',\n",
       "       'cluster_EM_PROBABILITY5', 'cluster_EM_PROBABILITY6',\n",
       "       'cluster_EM_PROBABILITY7', 'cluster_EM_PROBABILITY8',\n",
       "       'cluster_EM_PROBABILITY9', 'cluster_EM_PROBABILITY10',\n",
       "       'cluster_EM_PROBABILITY11', 'cluster_EM_PROBABILITY12',\n",
       "       'cluster_EM_PROBABILITY13', 'cluster_EM_PROBABILITY14',\n",
       "       'cluster_EM_PROBABILITY15', 'cluster_EM_PROBABILITY16',\n",
       "       'cluster_EM_PROBABILITY17', 'cluster_EM_PROBABILITY18',\n",
       "       'cluster_EM_PROBABILITY19', 'cluster_EM_PROBABILITY20',\n",
       "       'cluster_EM_PROBABILITY21', 'cluster_EM_PROBABILITY22',\n",
       "       'cluster_EM_PROBABILITY23', 'cluster_EM_PROBABILITY24',\n",
       "       'cluster_EM_PROBABILITY25', 'cluster_EM_PROBABILITY26',\n",
       "       'cluster_EM_PROBABILITY27', 'cluster_EM_PROBABILITY28',\n",
       "       'cluster_EM_PROBABILITY29', 'cluster_EM_PROBABILITY30',\n",
       "       'cluster_EM_PROBABILITY31', 'cluster_EM_PROBABILITY32',\n",
       "       'cluster_HAD_WEIGHT0', 'cluster_HAD_WEIGHT1',\n",
       "       'cluster_HAD_WEIGHT2', 'cluster_HAD_WEIGHT3',\n",
       "       'cluster_HAD_WEIGHT4', 'cluster_HAD_WEIGHT5',\n",
       "       'cluster_HAD_WEIGHT6', 'cluster_HAD_WEIGHT7',\n",
       "       'cluster_HAD_WEIGHT8', 'cluster_HAD_WEIGHT9',\n",
       "       'cluster_HAD_WEIGHT10', 'cluster_HAD_WEIGHT11',\n",
       "       'cluster_HAD_WEIGHT12', 'cluster_HAD_WEIGHT13',\n",
       "       'cluster_HAD_WEIGHT14', 'cluster_HAD_WEIGHT15',\n",
       "       'cluster_HAD_WEIGHT16', 'cluster_HAD_WEIGHT17',\n",
       "       'cluster_HAD_WEIGHT18', 'cluster_HAD_WEIGHT19',\n",
       "       'cluster_HAD_WEIGHT20', 'cluster_HAD_WEIGHT21',\n",
       "       'cluster_HAD_WEIGHT22', 'cluster_HAD_WEIGHT23',\n",
       "       'cluster_HAD_WEIGHT24', 'cluster_HAD_WEIGHT25',\n",
       "       'cluster_HAD_WEIGHT26', 'cluster_HAD_WEIGHT27',\n",
       "       'cluster_HAD_WEIGHT28', 'cluster_HAD_WEIGHT29',\n",
       "       'cluster_HAD_WEIGHT30', 'cluster_HAD_WEIGHT31',\n",
       "       'cluster_HAD_WEIGHT32', 'cluster_OOC_WEIGHT0',\n",
       "       'cluster_OOC_WEIGHT1', 'cluster_OOC_WEIGHT2',\n",
       "       'cluster_OOC_WEIGHT3', 'cluster_OOC_WEIGHT4',\n",
       "       'cluster_OOC_WEIGHT5', 'cluster_OOC_WEIGHT6',\n",
       "       'cluster_OOC_WEIGHT7', 'cluster_OOC_WEIGHT8',\n",
       "       'cluster_OOC_WEIGHT9', 'cluster_OOC_WEIGHT10',\n",
       "       'cluster_OOC_WEIGHT11', 'cluster_OOC_WEIGHT12',\n",
       "       'cluster_OOC_WEIGHT13', 'cluster_OOC_WEIGHT14',\n",
       "       'cluster_OOC_WEIGHT15', 'cluster_OOC_WEIGHT16',\n",
       "       'cluster_OOC_WEIGHT17', 'cluster_OOC_WEIGHT18',\n",
       "       'cluster_OOC_WEIGHT19', 'cluster_OOC_WEIGHT20',\n",
       "       'cluster_OOC_WEIGHT21', 'cluster_OOC_WEIGHT22',\n",
       "       'cluster_OOC_WEIGHT23', 'cluster_OOC_WEIGHT24',\n",
       "       'cluster_OOC_WEIGHT25', 'cluster_OOC_WEIGHT26',\n",
       "       'cluster_OOC_WEIGHT27', 'cluster_OOC_WEIGHT28',\n",
       "       'cluster_OOC_WEIGHT29', 'cluster_OOC_WEIGHT30',\n",
       "       'cluster_OOC_WEIGHT31', 'cluster_OOC_WEIGHT32',\n",
       "       'cluster_DM_WEIGHT0', 'cluster_DM_WEIGHT1', 'cluster_DM_WEIGHT2',\n",
       "       'cluster_DM_WEIGHT3', 'cluster_DM_WEIGHT4', 'cluster_DM_WEIGHT5',\n",
       "       'cluster_DM_WEIGHT6', 'cluster_DM_WEIGHT7', 'cluster_DM_WEIGHT8',\n",
       "       'cluster_DM_WEIGHT9', 'cluster_DM_WEIGHT10', 'cluster_DM_WEIGHT11',\n",
       "       'cluster_DM_WEIGHT12', 'cluster_DM_WEIGHT13',\n",
       "       'cluster_DM_WEIGHT14', 'cluster_DM_WEIGHT15',\n",
       "       'cluster_DM_WEIGHT16', 'cluster_DM_WEIGHT17',\n",
       "       'cluster_DM_WEIGHT18', 'cluster_DM_WEIGHT19',\n",
       "       'cluster_DM_WEIGHT20', 'cluster_DM_WEIGHT21',\n",
       "       'cluster_DM_WEIGHT22', 'cluster_DM_WEIGHT23',\n",
       "       'cluster_DM_WEIGHT24', 'cluster_DM_WEIGHT25',\n",
       "       'cluster_DM_WEIGHT26', 'cluster_DM_WEIGHT27',\n",
       "       'cluster_DM_WEIGHT28', 'cluster_DM_WEIGHT29',\n",
       "       'cluster_DM_WEIGHT30', 'cluster_DM_WEIGHT31',\n",
       "       'cluster_DM_WEIGHT32', 'cluster_CENTER_MAG0',\n",
       "       'cluster_CENTER_MAG1', 'cluster_CENTER_MAG2',\n",
       "       'cluster_CENTER_MAG3', 'cluster_CENTER_MAG4',\n",
       "       'cluster_CENTER_MAG5', 'cluster_CENTER_MAG6',\n",
       "       'cluster_CENTER_MAG7', 'cluster_CENTER_MAG8',\n",
       "       'cluster_CENTER_MAG9', 'cluster_CENTER_MAG10',\n",
       "       'cluster_CENTER_MAG11', 'cluster_CENTER_MAG12',\n",
       "       'cluster_CENTER_MAG13', 'cluster_CENTER_MAG14',\n",
       "       'cluster_CENTER_MAG15', 'cluster_CENTER_MAG16',\n",
       "       'cluster_CENTER_MAG17', 'cluster_CENTER_MAG18',\n",
       "       'cluster_CENTER_MAG19', 'cluster_CENTER_MAG20',\n",
       "       'cluster_CENTER_MAG21', 'cluster_CENTER_MAG22',\n",
       "       'cluster_CENTER_MAG23', 'cluster_CENTER_MAG24',\n",
       "       'cluster_CENTER_MAG25', 'cluster_CENTER_MAG26',\n",
       "       'cluster_CENTER_MAG27', 'cluster_CENTER_MAG28',\n",
       "       'cluster_CENTER_MAG29', 'cluster_CENTER_MAG30',\n",
       "       'cluster_CENTER_MAG31', 'cluster_CENTER_MAG32',\n",
       "       'cluster_FIRST_ENG_DENS0', 'cluster_FIRST_ENG_DENS1',\n",
       "       'cluster_FIRST_ENG_DENS2', 'cluster_FIRST_ENG_DENS3',\n",
       "       'cluster_FIRST_ENG_DENS4', 'cluster_FIRST_ENG_DENS5',\n",
       "       'cluster_FIRST_ENG_DENS6', 'cluster_FIRST_ENG_DENS7',\n",
       "       'cluster_FIRST_ENG_DENS8', 'cluster_FIRST_ENG_DENS9',\n",
       "       'cluster_FIRST_ENG_DENS10', 'cluster_FIRST_ENG_DENS11',\n",
       "       'cluster_FIRST_ENG_DENS12', 'cluster_FIRST_ENG_DENS13',\n",
       "       'cluster_FIRST_ENG_DENS14', 'cluster_FIRST_ENG_DENS15',\n",
       "       'cluster_FIRST_ENG_DENS16', 'cluster_FIRST_ENG_DENS17',\n",
       "       'cluster_FIRST_ENG_DENS18', 'cluster_FIRST_ENG_DENS19',\n",
       "       'cluster_FIRST_ENG_DENS20', 'cluster_FIRST_ENG_DENS21',\n",
       "       'cluster_FIRST_ENG_DENS22', 'cluster_FIRST_ENG_DENS23',\n",
       "       'cluster_FIRST_ENG_DENS24', 'cluster_FIRST_ENG_DENS25',\n",
       "       'cluster_FIRST_ENG_DENS26', 'cluster_FIRST_ENG_DENS27',\n",
       "       'cluster_FIRST_ENG_DENS28', 'cluster_FIRST_ENG_DENS29',\n",
       "       'cluster_FIRST_ENG_DENS30', 'cluster_FIRST_ENG_DENS31',\n",
       "       'cluster_FIRST_ENG_DENS32', 'cluster_CENTER_LAMBDA0',\n",
       "       'cluster_CENTER_LAMBDA1', 'cluster_CENTER_LAMBDA2',\n",
       "       'cluster_CENTER_LAMBDA3', 'cluster_CENTER_LAMBDA4',\n",
       "       'cluster_CENTER_LAMBDA5', 'cluster_CENTER_LAMBDA6',\n",
       "       'cluster_CENTER_LAMBDA7', 'cluster_CENTER_LAMBDA8',\n",
       "       'cluster_CENTER_LAMBDA9', 'cluster_CENTER_LAMBDA10',\n",
       "       'cluster_CENTER_LAMBDA11', 'cluster_CENTER_LAMBDA12',\n",
       "       'cluster_CENTER_LAMBDA13', 'cluster_CENTER_LAMBDA14',\n",
       "       'cluster_CENTER_LAMBDA15', 'cluster_CENTER_LAMBDA16',\n",
       "       'cluster_CENTER_LAMBDA17', 'cluster_CENTER_LAMBDA18',\n",
       "       'cluster_CENTER_LAMBDA19', 'cluster_CENTER_LAMBDA20',\n",
       "       'cluster_CENTER_LAMBDA21', 'cluster_CENTER_LAMBDA22',\n",
       "       'cluster_CENTER_LAMBDA23', 'cluster_CENTER_LAMBDA24',\n",
       "       'cluster_CENTER_LAMBDA25', 'cluster_CENTER_LAMBDA26',\n",
       "       'cluster_CENTER_LAMBDA27', 'cluster_CENTER_LAMBDA28',\n",
       "       'cluster_CENTER_LAMBDA29', 'cluster_CENTER_LAMBDA30',\n",
       "       'cluster_CENTER_LAMBDA31', 'cluster_CENTER_LAMBDA32',\n",
       "       'cluster_ISOLATION0', 'cluster_ISOLATION1', 'cluster_ISOLATION2',\n",
       "       'cluster_ISOLATION3', 'cluster_ISOLATION4', 'cluster_ISOLATION5',\n",
       "       'cluster_ISOLATION6', 'cluster_ISOLATION7', 'cluster_ISOLATION8',\n",
       "       'cluster_ISOLATION9', 'cluster_ISOLATION10', 'cluster_ISOLATION11',\n",
       "       'cluster_ISOLATION12', 'cluster_ISOLATION13',\n",
       "       'cluster_ISOLATION14', 'cluster_ISOLATION15',\n",
       "       'cluster_ISOLATION16', 'cluster_ISOLATION17',\n",
       "       'cluster_ISOLATION18', 'cluster_ISOLATION19',\n",
       "       'cluster_ISOLATION20', 'cluster_ISOLATION21',\n",
       "       'cluster_ISOLATION22', 'cluster_ISOLATION23',\n",
       "       'cluster_ISOLATION24', 'cluster_ISOLATION25',\n",
       "       'cluster_ISOLATION26', 'cluster_ISOLATION27',\n",
       "       'cluster_ISOLATION28', 'cluster_ISOLATION29',\n",
       "       'cluster_ISOLATION30', 'cluster_ISOLATION31',\n",
       "       'cluster_ISOLATION32', 'trackPt', 'trackP', 'trackMass',\n",
       "       'trackEta', 'trackPhi', 'trackNumberOfPixelHits',\n",
       "       'trackNumberOfSCTHits', 'trackNumberOfPixelDeadSensors',\n",
       "       'trackNumberOfSCTDeadSensors',\n",
       "       'trackNumberOfInnermostPixelLayerHits',\n",
       "       'trackNumberOfNextToInnermostPixelLayerHits',\n",
       "       'trackExpectInnermostPixelLayerHit',\n",
       "       'trackExpectNextToInnermostPixelLayerHit', 'trackNumberOfTRTHits',\n",
       "       'trackNumberOfTRTOutliers', 'trackChiSquared', 'trackNumberDOF',\n",
       "       'trackD0', 'trackZ0', 'track_pt', 'track_eta', 'track_phi',\n",
       "       'track_z0', 'truth_particle_e', 'log10_cluster_e_0',\n",
       "       'log10_cluster_e_1', 'log10_cluster_e_2', 'log10_cluster_e_3',\n",
       "       'log10_cluster_e_4', 'log10_cluster_e_5', 'log10_cluster_e_6',\n",
       "       'log10_cluster_e_7', 'log10_cluster_e_8', 'log10_cluster_e_9',\n",
       "       'log10_cluster_e_10', 'log10_cluster_e_11', 'log10_cluster_e_12',\n",
       "       'log10_cluster_e_13', 'log10_cluster_e_14', 'log10_cluster_e_15',\n",
       "       'log10_cluster_e_16', 'log10_cluster_e_17', 'log10_cluster_e_18',\n",
       "       'log10_cluster_e_19', 'log10_cluster_e_20', 'log10_cluster_e_21',\n",
       "       'log10_cluster_e_22', 'log10_cluster_e_23', 'log10_cluster_e_24',\n",
       "       'log10_cluster_e_25', 'log10_cluster_e_26', 'log10_cluster_e_27',\n",
       "       'log10_cluster_e_28', 'log10_cluster_e_29', 'log10_cluster_e_30',\n",
       "       'log10_cluster_e_31', 'log10_cluster_e_32', 'log10_track_pt',\n",
       "       'log10_truth_particle_e'], dtype=object)"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array(df_train.keys())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc1de1b0-b31a-4287-a359-18b83f382c55",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Preprocess"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "id": "b774ef68-b13a-42f3-a55c-0045c538f41f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:08.839979Z",
     "iopub.status.busy": "2022-08-09T08:08:08.839703Z",
     "iopub.status.idle": "2022-08-09T08:08:08.862716Z",
     "shell.execute_reply": "2022-08-09T08:08:08.861827Z",
     "shell.execute_reply.started": "2022-08-09T08:08:08.839949Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['log10_cluster_e_0', 'log10_cluster_e_1', 'log10_cluster_e_2', 'log10_cluster_e_3', 'log10_cluster_e_4', 'log10_cluster_e_5', 'log10_cluster_e_6', 'log10_cluster_e_7', 'log10_cluster_e_8', 'log10_cluster_e_9', 'log10_cluster_e_10', 'log10_cluster_e_11', 'log10_cluster_e_12', 'log10_cluster_e_13', 'log10_cluster_e_14', 'log10_cluster_e_15', 'log10_cluster_e_16', 'log10_cluster_e_17', 'log10_cluster_e_18', 'log10_cluster_e_19', 'log10_cluster_e_20', 'log10_cluster_e_21', 'log10_cluster_e_22', 'log10_cluster_e_23', 'log10_cluster_e_24', 'log10_cluster_e_25', 'log10_cluster_e_26', 'log10_cluster_e_27', 'log10_cluster_e_28', 'log10_track_pt', 'track_eta', 'track_phi', 'track_z0']\n"
     ]
    }
   ],
   "source": [
    "### All clusters + track pt/eta/phi/z0\n",
    "train_vars = [var for var in df_train.keys() if var.startswith('log10') and var != 'log10_truth_particle_e']\n",
    "train_vars += ['track_eta', 'track_phi', 'track_z0']\n",
    "# train_vars += [var for var in df_train.keys() if var.startswith('track')]\n",
    "# train_vars += [var for var in df_train.keys() if var.startswith('cluster') and not var.startswith('cluster_e')]\n",
    "\n",
    "for i in range(29, 33):\n",
    "    train_vars.remove('log10_cluster_e_'+str(i))\n",
    "    # for var in cluster_var:\n",
    "    #     train_vars.remove(var+str(i))\n",
    "\n",
    "print(train_vars)\n",
    "\n",
    "### Augment df_test with any extra columns needed\n",
    "df_test[[var for var in df_train.keys() if var not in df_test.keys()]] = 0 \n",
    "\n",
    "train_x = df_train[train_vars].values\n",
    "train_y = df_train['log10_truth_particle_e'].values\n",
    "val_x = df_val[train_vars].values\n",
    "val_y = df_val['log10_truth_particle_e'].values\n",
    "test_x = df_test[train_vars].values\n",
    "test_y = df_test['log10_truth_particle_e'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "id": "83ef2fcb-3028-4408-8057-367e779be53b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:11.424747Z",
     "iopub.status.busy": "2022-08-09T08:08:11.424483Z",
     "iopub.status.idle": "2022-08-09T08:08:11.535663Z",
     "shell.execute_reply": "2022-08-09T08:08:11.534874Z",
     "shell.execute_reply.started": "2022-08-09T08:08:11.424718Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "### Normalize the inputs \n",
    "sc = StandardScaler()\n",
    "train_x = sc.fit_transform(train_x)\n",
    "val_x = sc.transform(val_x)\n",
    "test_x = sc.transform(test_x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "id": "804df3cf-042c-4b8e-90bb-477e8f9c38df",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:11.605494Z",
     "iopub.status.busy": "2022-08-09T08:08:11.605374Z",
     "iopub.status.idle": "2022-08-09T08:08:11.609507Z",
     "shell.execute_reply": "2022-08-09T08:08:11.609012Z",
     "shell.execute_reply.started": "2022-08-09T08:08:11.605480Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "33"
      ]
     },
     "execution_count": 220,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_x.shape[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "id": "111bffb9-daee-4a98-95cd-73938df8556f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:12.895978Z",
     "iopub.status.busy": "2022-08-09T08:08:12.895856Z",
     "iopub.status.idle": "2022-08-09T08:08:12.900019Z",
     "shell.execute_reply": "2022-08-09T08:08:12.899492Z",
     "shell.execute_reply.started": "2022-08-09T08:08:12.895965Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['log10_cluster_e_0',\n",
       " 'log10_cluster_e_1',\n",
       " 'log10_cluster_e_2',\n",
       " 'log10_cluster_e_3',\n",
       " 'log10_cluster_e_4',\n",
       " 'log10_cluster_e_5',\n",
       " 'log10_cluster_e_6',\n",
       " 'log10_cluster_e_7',\n",
       " 'log10_cluster_e_8',\n",
       " 'log10_cluster_e_9',\n",
       " 'log10_cluster_e_10',\n",
       " 'log10_cluster_e_11',\n",
       " 'log10_cluster_e_12',\n",
       " 'log10_cluster_e_13',\n",
       " 'log10_cluster_e_14',\n",
       " 'log10_cluster_e_15',\n",
       " 'log10_cluster_e_16',\n",
       " 'log10_cluster_e_17',\n",
       " 'log10_cluster_e_18',\n",
       " 'log10_cluster_e_19',\n",
       " 'log10_cluster_e_20',\n",
       " 'log10_cluster_e_21',\n",
       " 'log10_cluster_e_22',\n",
       " 'log10_cluster_e_23',\n",
       " 'log10_cluster_e_24',\n",
       " 'log10_cluster_e_25',\n",
       " 'log10_cluster_e_26',\n",
       " 'log10_cluster_e_27',\n",
       " 'log10_cluster_e_28',\n",
       " 'log10_track_pt',\n",
       " 'track_eta',\n",
       " 'track_phi',\n",
       " 'track_z0']"
      ]
     },
     "execution_count": 221,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_vars"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "id": "0f258a4d-a1e5-486e-a76f-498c45e38917",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:14.960161Z",
     "iopub.status.busy": "2022-08-09T08:08:14.960034Z",
     "iopub.status.idle": "2022-08-09T08:08:14.964107Z",
     "shell.execute_reply": "2022-08-09T08:08:14.963561Z",
     "shell.execute_reply.started": "2022-08-09T08:08:14.960147Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def regression_model():\n",
    "    model = Sequential()\n",
    "    model.add(Dense(50, \n",
    "                    input_dim=train_x.shape[1], \n",
    "                    activation='relu'))\n",
    "    model.add(Dense(50, activation='relu'))\n",
    "    model.add(Dense(50, activation='relu'))\n",
    "    model.add(Dense(1))\n",
    "    model.compile(loss='mean_absolute_error', optimizer='adam')\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "id": "c7cc8b57-aa68-41c8-8a97-d512fc20e6f3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:15.247685Z",
     "iopub.status.busy": "2022-08-09T08:08:15.247415Z",
     "iopub.status.idle": "2022-08-09T08:08:15.273852Z",
     "shell.execute_reply": "2022-08-09T08:08:15.273104Z",
     "shell.execute_reply.started": "2022-08-09T08:08:15.247670Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_8\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_32 (Dense)             (None, 50)                1700      \n",
      "_________________________________________________________________\n",
      "dense_33 (Dense)             (None, 50)                2550      \n",
      "_________________________________________________________________\n",
      "dense_34 (Dense)             (None, 50)                2550      \n",
      "_________________________________________________________________\n",
      "dense_35 (Dense)             (None, 1)                 51        \n",
      "=================================================================\n",
      "Total params: 6,851\n",
      "Trainable params: 6,851\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = regression_model()\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fa176893-fe41-429d-8b05-fbae443e053f",
   "metadata": {},
   "source": [
    "# Load pre-trained weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "id": "4847a313-ae6e-43df-8cc8-da8387c4e698",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:17.300699Z",
     "iopub.status.busy": "2022-08-09T08:08:17.300433Z",
     "iopub.status.idle": "2022-08-09T08:08:17.303818Z",
     "shell.execute_reply": "2022-08-09T08:08:17.303132Z",
     "shell.execute_reply.started": "2022-08-09T08:08:17.300679Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# model.load_weights(\"dnn_best_weights_50.h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "08237162-481d-49ff-9740-d49755316e84",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Or, re-train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "id": "4d197211-348b-4017-9eb9-4c7e36cbd94c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:08:22.799472Z",
     "iopub.status.busy": "2022-08-09T08:08:22.799215Z",
     "iopub.status.idle": "2022-08-09T08:26:17.232527Z",
     "shell.execute_reply": "2022-08-09T08:26:17.231682Z",
     "shell.execute_reply.started": "2022-08-09T08:08:22.799453Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0558 - val_loss: 0.0335\n",
      "Epoch 2/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0322 - val_loss: 0.0317\n",
      "Epoch 3/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0293 - val_loss: 0.0303\n",
      "Epoch 4/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0278 - val_loss: 0.0268\n",
      "Epoch 5/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0269 - val_loss: 0.0254\n",
      "Epoch 6/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0263 - val_loss: 0.0292\n",
      "Epoch 7/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0259 - val_loss: 0.0272\n",
      "Epoch 8/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0255 - val_loss: 0.0240\n",
      "Epoch 9/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0254 - val_loss: 0.0253\n",
      "Epoch 10/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0252 - val_loss: 0.0235\n",
      "Epoch 11/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0249 - val_loss: 0.0241\n",
      "Epoch 12/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0246 - val_loss: 0.0258\n",
      "Epoch 13/50\n",
      "9422/9422 [==============================] - 20s 2ms/step - loss: 0.0245 - val_loss: 0.0244\n",
      "Epoch 14/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0243 - val_loss: 0.0240\n",
      "Epoch 15/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0240 - val_loss: 0.0235\n",
      "Epoch 16/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0239 - val_loss: 0.0233\n",
      "Epoch 17/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0238 - val_loss: 0.0252\n",
      "Epoch 18/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0238 - val_loss: 0.0236\n",
      "Epoch 19/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0238 - val_loss: 0.0230\n",
      "Epoch 20/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0237 - val_loss: 0.0231\n",
      "Epoch 21/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0235 - val_loss: 0.0240\n",
      "Epoch 22/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0235 - val_loss: 0.0249\n",
      "Epoch 23/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0235 - val_loss: 0.0228\n",
      "Epoch 24/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0235 - val_loss: 0.0238\n",
      "Epoch 25/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0233 - val_loss: 0.0230\n",
      "Epoch 26/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0232 - val_loss: 0.0234\n",
      "Epoch 27/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0232 - val_loss: 0.0226\n",
      "Epoch 28/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0231 - val_loss: 0.0243\n",
      "Epoch 29/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0230 - val_loss: 0.0232\n",
      "Epoch 30/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0232 - val_loss: 0.0223\n",
      "Epoch 31/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0231 - val_loss: 0.0238\n",
      "Epoch 32/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0230 - val_loss: 0.0229\n",
      "Epoch 33/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0230 - val_loss: 0.0226\n",
      "Epoch 34/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0229 - val_loss: 0.0240\n",
      "Epoch 35/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0229 - val_loss: 0.0227\n",
      "Epoch 36/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0230 - val_loss: 0.0226\n",
      "Epoch 37/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0230 - val_loss: 0.0237\n",
      "Epoch 38/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0228 - val_loss: 0.0226\n",
      "Epoch 39/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0227 - val_loss: 0.0222\n",
      "Epoch 40/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0228 - val_loss: 0.0222\n",
      "Epoch 41/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0227 - val_loss: 0.0227\n",
      "Epoch 42/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0227 - val_loss: 0.0231\n",
      "Epoch 43/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0227 - val_loss: 0.0227\n",
      "Epoch 44/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0227 - val_loss: 0.0237\n",
      "Epoch 45/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0227 - val_loss: 0.0223\n",
      "Epoch 46/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0226 - val_loss: 0.0223\n",
      "Epoch 47/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0226 - val_loss: 0.0224\n",
      "Epoch 48/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0225 - val_loss: 0.0228\n",
      "Epoch 49/50\n",
      "9422/9422 [==============================] - 22s 2ms/step - loss: 0.0225 - val_loss: 0.0224\n",
      "Epoch 50/50\n",
      "9422/9422 [==============================] - 21s 2ms/step - loss: 0.0225 - val_loss: 0.0223\n",
      "CPU times: user 26min 55s, sys: 2min 50s, total: 29min 46s\n",
      "Wall time: 17min 54s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "history = model.fit(\n",
    "    train_x,\n",
    "    train_y,\n",
    "    validation_data=(val_x, val_y),\n",
    "#     validation_split=0.2,\n",
    "    batch_size=32,\n",
    "    verbose=1, epochs=50)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8703b080-e98c-4b1b-b989-47239d461479",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Evaluate on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "id": "a476bd2d-1e1a-4109-848a-b79679354118",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:26:36.482325Z",
     "iopub.status.busy": "2022-08-09T08:26:36.481992Z",
     "iopub.status.idle": "2022-08-09T08:26:38.924118Z",
     "shell.execute_reply": "2022-08-09T08:26:38.923551Z",
     "shell.execute_reply.started": "2022-08-09T08:26:36.482289Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'Predicted Energy / True Energy')"
      ]
     },
     "execution_count": 230,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1200x800 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Evaluate performance on test set \n",
    "plt.figure(dpi=200)\n",
    "df_test['nn_output'] = model.predict(test_x)\n",
    "plt.hist(10**df_test.nn_output/10**df_test.log10_truth_particle_e, bins=np.linspace(0.,2.,40));\n",
    "plt.yscale('log')\n",
    "plt.xlabel('Predicted Energy / True Energy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "id": "08ab5b3c-33a0-4155-b617-4b9d44c08a06",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:26:38.925174Z",
     "iopub.status.busy": "2022-08-09T08:26:38.925059Z",
     "iopub.status.idle": "2022-08-09T08:26:38.991634Z",
     "shell.execute_reply": "2022-08-09T08:26:38.991092Z",
     "shell.execute_reply.started": "2022-08-09T08:26:38.925160Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.018506587308599238"
      ]
     },
     "execution_count": 231,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.min(10**df_test[df_test.truth_particle_e < 100].nn_output/10**df_test[df_test.truth_particle_e < 100].log10_truth_particle_e)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "id": "b8d40b54-0480-45b1-b48e-ec6b280d787a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:26:38.992465Z",
     "iopub.status.busy": "2022-08-09T08:26:38.992353Z",
     "iopub.status.idle": "2022-08-09T08:26:39.421968Z",
     "shell.execute_reply": "2022-08-09T08:26:39.421386Z",
     "shell.execute_reply.started": "2022-08-09T08:26:38.992451Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2400x1600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Response median plot \n",
    "import scipy.stats as stats\n",
    "from matplotlib.colors import ListedColormap\n",
    "from matplotlib.colors import LogNorm\n",
    "\n",
    "x = df_test.truth_particle_e\n",
    "y = (10**df_test.nn_output)/x\n",
    "\n",
    "xbin = [10**exp for exp in np.arange(-1., 3.1, 0.05)]\n",
    "ybin = np.arange(0., 3.1, 0.05)\n",
    "xcenter = [(xbin[i] + xbin[i+1]) / 2 for i in range(len(xbin)-1)]\n",
    "profileXMed = stats.binned_statistic(\n",
    "    x, y, bins=xbin, statistic='median').statistic\n",
    "\n",
    "c_map = ListedColormap(sns.color_palette(\"Blues\", n_colors=100).as_hex())\n",
    "fig = plt.figure(figsize=(12,8), dpi=200)\n",
    "fig.patch.set_facecolor('white')\n",
    "plt.hist2d(x, y, bins=[xbin, ybin], norm=LogNorm(),zorder = -1, cmap=c_map);\n",
    "plt.plot(np.array(xcenter), np.array(profileXMed), color='indianred', linewidth=3)\n",
    "plt.plot([0.1, 1000], [1, 1], linestyle='--', color='black');\n",
    "plt.xscale('log')\n",
    "plt.ylim(0, 1.75)\n",
    "plt.xlim(0.3, )\n",
    "plt.xlabel('Truth Particle Energy [GeV]')\n",
    "plt.ylabel('Predicted Energy / Target');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "id": "8c39aab3-5fbf-430f-bfb5-162bd8efd9c2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-08-09T08:26:39.423188Z",
     "iopub.status.busy": "2022-08-09T08:26:39.422977Z",
     "iopub.status.idle": "2022-08-09T08:26:39.792487Z",
     "shell.execute_reply": "2022-08-09T08:26:39.791907Z",
     "shell.execute_reply.started": "2022-08-09T08:26:39.423172Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2000x1200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### IQR plot \n",
    "\n",
    "def iqrOverMed(x):\n",
    "    # get the IQR via the percentile function\n",
    "    # 84 is median + 1 sigma, 16 is median - 1 sigma\n",
    "    q84, q16 = np.percentile(x, [84, 16])\n",
    "    iqr = q84 - q16\n",
    "    med = np.median(x)\n",
    "    return iqr / (2*med)\n",
    "\n",
    "xbin = [10**exp for exp in np.arange(-1., 3.1, 0.1)]\n",
    "xcenter = [(xbin[i] + xbin[i+1]) / 2 for i in range(len(xbin)-1)]\n",
    "resolution = stats.binned_statistic(x, y, bins=xbin, statistic=iqrOverMed).statistic\n",
    "\n",
    "fig = plt.figure(figsize=(10,6), dpi=200)\n",
    "fig.patch.set_facecolor('white')\n",
    "plt.plot(xcenter, resolution, linewidth=3)\n",
    "plt.xscale('log')\n",
    "plt.xlim(0.1, 1000)\n",
    "plt.ylim(0,0.5)\n",
    "plt.xlabel('Truth Particle Energy [GeV]')\n",
    "plt.ylabel('Response IQR / 2 x Median');"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc3c8088-7256-4f51-a214-c28fce42b53e",
   "metadata": {},
   "source": [
    "Compare with other models..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f8d44fcd-b3b6-4fa1-9a98-e84a51f05d08",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pred_file = 'results/onetrack_multicluster/piyush_june22/gnn_allclusters_cells/inference_predictions.npz'\n",
    "predictions = np.load(pred_file)\n",
    "targets_regress = predictions['targets']\n",
    "outputs_regress = predictions['outputs'][:,0]\n",
    "df = pd.DataFrame(np.vstack([targets_regress, outputs_regress]).T, columns=['targets_regress', 'outputs_regress'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6f652d02-7635-4551-9cb1-de54e148e8c5",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "plt.figure(dpi=200)\n",
    "plt.hist(df_transformer.outputs_regress/df_transformer.targets_regress, \n",
    "        bins=np.linspace(0.5,1.5,100),\n",
    "         label=\"Transformer\",\n",
    "        );\n",
    "plt.legend()\n",
    "plt.xlabel(\"Predicted Energy / Target Energy\");\n",
    "plt.ylabel(\"Events\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "adbdb8c3-5b0c-4328-9ede-43aeaeffd412",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "plt.figure(dpi=200)\n",
    "plt.hist(df_test.truth_particle_e, label=\"DNN\", histtype=\"step\", alpha=1., density=True, linewidth=2);\n",
    "plt.hist(df.targets_regress, label=\"GNN\", histtype=\"step\", alpha=1., density=True, linewidth=2);\n",
    "plt.hist(df_transformer.targets_regress, label=\"Transformer\", histtype=\"step\", alpha=1., density=True, linewidth=2);\n",
    "plt.legend();\n",
    "plt.xlabel(\"Target Energy [GeV]\");"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6250c5dc-1d49-44aa-8b9f-b396fa32b866",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true,
    "tags": []
   },
   "source": [
    "# Misc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b4a9f654-a6c3-4d38-ab6b-a8195e56c7c9",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.read_hdf(\"train_dnn.h5\")\n",
    "val = pd.read_hdf(\"val_dnn.h5\")\n",
    "test = pd.read_hdf(\"test_dnn.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bc7d7910-ab96-41ad-85af-48e057d5f756",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import uproot\n",
    "from glob import glob\n",
    "import json \n",
    "import os\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "pi0_dir = '/clusterfs/ml4hep/mpettee/ml4pions/data/root_files/user.angerami.mc16_13TeV.900246.PG_singlepi0_logE0p2to2000.e8312_e7400_s3170_r12383.v01-45-gaa27bcb_OutputStream/'\n",
    "pion_dir = '/clusterfs/ml4hep/mpettee/ml4pions/data/root_files/user.angerami.mc16_13TeV.900247.PG_singlepion_logE0p2to2000.e8312_e7400_s3170_r12383.v01-45-gaa27bcb_OutputStream/'\n",
    "pi0_files = sorted(glob(pi0_dir+\"*.root\"))\n",
    "pion_files = sorted(glob(pion_dir+\"*.root\"))\n",
    "\n",
    "variables = [\"nTrack\", \"truthPartE\"]\n",
    "\n",
    "df_all = pd.concat([pd.DataFrame(uproot.open(file)[\"EventTree\"].arrays(variables, library = \"np\")) \n",
    "                for file in tqdm(pion_files)])\n",
    "df_all[\"truthPartE\"] = df_all.truthPartE.astype('float')\n",
    "df_all.reset_index(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "17f104a7-645f-47ad-9a10-d823811776a1",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pi0_dir = '/clusterfs/ml4hep/mpettee/ml4pions/data/root_files/user.angerami.mc16_13TeV.900246.PG_singlepi0_logE0p2to2000.e8312_e7400_s3170_r12383.v01-45-gaa27bcb_OutputStream/'\n",
    "pion_dir = '/clusterfs/ml4hep/mpettee/ml4pions/data/root_files/user.angerami.mc16_13TeV.900247.PG_singlepion_logE0p2to2000.e8312_e7400_s3170_r12383.v01-45-gaa27bcb_OutputStream/'\n",
    "pi0_files = sorted(glob(pi0_dir+\"*.root\"))\n",
    "pion_files = sorted(glob(pion_dir+\"*.root\"))\n",
    "\n",
    "variables = [\"nTrack\", \"truthPartE\", \"cluster_E\"]\n",
    "\n",
    "df = pd.concat([pd.DataFrame(uproot.open(file)[\"EventTree\"].arrays(variables, library = \"np\")) \n",
    "                for file in tqdm(pion_files)])\n",
    "df[\"truthPartE\"] = df.truthPartE.astype('float')\n",
    "df.reset_index(inplace=True)\n",
    "\n",
    "print(df[(df.truthPartE < 0.201)].nTrack.value_counts())\n",
    "print(\"Minimum truthPartE with nTrack == 1 is {:.5f}.\".format(df[df.nTrack == 1].truthPartE.min()))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "97fd6504-d635-4286-815f-6275a13cd73f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "cluster_e = []\n",
    "for i in tqdm(range(len(df))): \n",
    "    event_pass = True\n",
    "    for cluster in range(len(df.cluster_E.iloc[i])):\n",
    "        if np.array(df.cluster_E.iloc[i][cluster]) > 0.5: \n",
    "            event_pass = False\n",
    "            continue\n",
    "    cluster_e.append(event_pass)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "nbdev",
   "language": "python",
   "name": "nbdev"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}