misc/example/00_model_example_esol_descriptors.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "from molmap.model import RegressionEstimator, MultiClassEstimator, MultiLabelEstimator\n",
    "from sklearn.preprocessing import StandardScaler, MinMaxScaler\n",
    "from chembench import dataset\n",
    "from sklearn.utils import shuffle \n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from molmap import MolMap\n",
    "\n",
    "def Rdsplit(df, random_state = 888, split_size = [0.8, 0.1, 0.1]):\n",
    "    base_indices = np.arange(len(df)) \n",
    "    base_indices = shuffle(base_indices, random_state = random_state) \n",
    "    nb_test = int(len(base_indices) * split_size[2]) \n",
    "    nb_val = int(len(base_indices) * split_size[1]) \n",
    "    test_idx = base_indices[0:nb_test] \n",
    "    valid_idx = base_indices[(nb_test):(nb_test+nb_val)] \n",
    "    train_idx = base_indices[(nb_test+nb_val):len(base_indices)] \n",
    "    print(len(train_idx), len(valid_idx), len(test_idx)) \n",
    "    return train_idx, valid_idx, test_idx "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total samples: 1128\n"
     ]
    }
   ],
   "source": [
    "data = dataset.load_ESOL()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "mp = MolMap(ftype='descriptor',metric='cosine',)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2021-03-31 10:29:01,510 - \u001b[32mINFO\u001b[0m - [bidd-molmap]\u001b[0m - Applying grid feature map(assignment), this may take several minutes(1~30 min)\u001b[0m\n",
      "2021-03-31 10:29:03,927 - \u001b[32mINFO\u001b[0m - [bidd-molmap]\u001b[0m - Finished\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "mp.fit(verbose=0, method='umap', min_dist=0.1, n_neighbors=15,)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|##########| 1128/1128 [04:46<00:00,  2.90it/s]\n"
     ]
    }
   ],
   "source": [
    "X = mp.batch_transform(data.x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y = data.y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "904 112 112\n"
     ]
    }
   ],
   "source": [
    "train_idx, valid_idx, test_idx  = Rdsplit(data.df, random_state = 888)\n",
    "trainX = X[train_idx]\n",
    "trainY = Y[train_idx]\n",
    "validX = X[valid_idx]\n",
    "validY = Y[valid_idx]\n",
    "testX = X[test_idx]\n",
    "testY = Y[test_idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RegressionEstimator(batch_size=8, gpuid='1', patience=20)\n"
     ]
    }
   ],
   "source": [
    "# define your model\n",
    "clf = RegressionEstimator(n_outputs=trainY.shape[1], \n",
    "                          fmap_shape1 = trainX.shape[1:], \n",
    "                          dense_layers = [128, 64],  \n",
    "                          batch_size = 8,\n",
    "                          y_scale= None, \n",
    "                          patience = 20, \n",
    "                          gpuid = 1) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch: 0001, loss: 5.4672 - val_loss: 3.8865; rmse: 2.0240 - rmse_val: 1.9714;  r2: 0.4001 - r2_val: 0.3690                                                                                                    \n",
      "epoch: 0002, loss: 3.8308 - val_loss: 3.2697; rmse: 1.8406 - rmse_val: 1.8082;  r2: 0.4249 - r2_val: 0.3721                                                                                                    \n",
      "epoch: 0003, loss: 3.0293 - val_loss: 2.5779; rmse: 1.6007 - rmse_val: 1.6056;  r2: 0.5106 - r2_val: 0.4504                                                                                                    \n",
      "epoch: 0004, loss: 2.1995 - val_loss: 2.2505; rmse: 1.4692 - rmse_val: 1.5002;  r2: 0.6367 - r2_val: 0.5689                                                                                                    \n",
      "epoch: 0005, loss: 1.7716 - val_loss: 2.1318; rmse: 1.4421 - rmse_val: 1.4601;  r2: 0.6673 - r2_val: 0.6430                                                                                                    \n",
      "epoch: 0006, loss: 1.4260 - val_loss: 1.3212; rmse: 1.1045 - rmse_val: 1.1494;  r2: 0.7373 - r2_val: 0.6994                                                                                                    \n",
      "epoch: 0007, loss: 1.2364 - val_loss: 1.1180; rmse: 1.0184 - rmse_val: 1.0573;  r2: 0.7676 - r2_val: 0.7319                                                                                                    \n",
      "epoch: 0008, loss: 1.0656 - val_loss: 1.0171; rmse: 0.9743 - rmse_val: 1.0085;  r2: 0.7945 - r2_val: 0.7578                                                                                                    \n",
      "epoch: 0009, loss: 0.9795 - val_loss: 0.8923; rmse: 0.9036 - rmse_val: 0.9446;  r2: 0.8169 - r2_val: 0.7881                                                                                                    \n",
      "epoch: 0010, loss: 0.8879 - val_loss: 0.8051; rmse: 0.8724 - rmse_val: 0.8973;  r2: 0.8371 - r2_val: 0.8085                                                                                                    \n",
      "epoch: 0011, loss: 0.8090 - val_loss: 0.7407; rmse: 0.8153 - rmse_val: 0.8606;  r2: 0.8559 - r2_val: 0.8317                                                                                                    \n",
      "epoch: 0012, loss: 0.7135 - val_loss: 0.7006; rmse: 0.8101 - rmse_val: 0.8370;  r2: 0.8633 - r2_val: 0.8360                                                                                                    \n",
      "epoch: 0013, loss: 0.6760 - val_loss: 0.6657; rmse: 0.7980 - rmse_val: 0.8159;  r2: 0.8786 - r2_val: 0.8513                                                                                                    \n",
      "epoch: 0014, loss: 0.6842 - val_loss: 0.7868; rmse: 0.8254 - rmse_val: 0.8870;  r2: 0.8819 - r2_val: 0.8673                                                                                                    \n",
      "epoch: 0015, loss: 0.5755 - val_loss: 0.5500; rmse: 0.7410 - rmse_val: 0.7417;  r2: 0.8917 - r2_val: 0.8727                                                                                                    \n",
      "epoch: 0016, loss: 0.5599 - val_loss: 0.5884; rmse: 0.7730 - rmse_val: 0.7671;  r2: 0.8985 - r2_val: 0.8789                                                                                                    \n",
      "epoch: 0017, loss: 0.5208 - val_loss: 0.5388; rmse: 0.7266 - rmse_val: 0.7340;  r2: 0.9025 - r2_val: 0.8824                                                                                                    \n",
      "epoch: 0018, loss: 0.5014 - val_loss: 0.4450; rmse: 0.6368 - rmse_val: 0.6671;  r2: 0.9099 - r2_val: 0.8956                                                                                                    \n",
      "epoch: 0019, loss: 0.5047 - val_loss: 0.5922; rmse: 0.6938 - rmse_val: 0.7695;  r2: 0.9133 - r2_val: 0.8954                                                                                                    \n",
      "epoch: 0020, loss: 0.4316 - val_loss: 0.4294; rmse: 0.6188 - rmse_val: 0.6553;  r2: 0.9164 - r2_val: 0.9044                                                                                                    \n",
      "epoch: 0021, loss: 0.4202 - val_loss: 0.4128; rmse: 0.5983 - rmse_val: 0.6425;  r2: 0.9220 - r2_val: 0.9067                                                                                                    \n",
      "epoch: 0022, loss: 0.4167 - val_loss: 0.3653; rmse: 0.5815 - rmse_val: 0.6044;  r2: 0.9247 - r2_val: 0.9135                                                                                                    \n",
      "epoch: 0023, loss: 0.4235 - val_loss: 0.4396; rmse: 0.6420 - rmse_val: 0.6630;  r2: 0.9246 - r2_val: 0.9038                                                                                                    \n",
      "epoch: 0024, loss: 0.3640 - val_loss: 0.3677; rmse: 0.5655 - rmse_val: 0.6064;  r2: 0.9309 - r2_val: 0.9152                                                                                                    \n",
      "epoch: 0025, loss: 0.4095 - val_loss: 0.7709; rmse: 0.8826 - rmse_val: 0.8780;  r2: 0.9297 - r2_val: 0.9071                                                                                                    \n",
      "epoch: 0026, loss: 0.4005 - val_loss: 0.3591; rmse: 0.5646 - rmse_val: 0.5993;  r2: 0.9309 - r2_val: 0.9143                                                                                                    \n",
      "epoch: 0027, loss: 0.3768 - val_loss: 0.3993; rmse: 0.5625 - rmse_val: 0.6319;  r2: 0.9347 - r2_val: 0.9163                                                                                                    \n",
      "epoch: 0028, loss: 0.3278 - val_loss: 0.3509; rmse: 0.5418 - rmse_val: 0.5924;  r2: 0.9369 - r2_val: 0.9165                                                                                                    \n",
      "epoch: 0029, loss: 0.3659 - val_loss: 0.4263; rmse: 0.5883 - rmse_val: 0.6529;  r2: 0.9374 - r2_val: 0.9239                                                                                                    \n",
      "epoch: 0030, loss: 0.3679 - val_loss: 0.3388; rmse: 0.5674 - rmse_val: 0.5821;  r2: 0.9386 - r2_val: 0.9235                                                                                                    \n",
      "epoch: 0031, loss: 0.3522 - val_loss: 0.3470; rmse: 0.5314 - rmse_val: 0.5891;  r2: 0.9418 - r2_val: 0.9187                                                                                                    \n",
      "epoch: 0032, loss: 0.3271 - val_loss: 0.3388; rmse: 0.5359 - rmse_val: 0.5820;  r2: 0.9437 - r2_val: 0.9232                                                                                                    \n",
      "epoch: 0033, loss: 0.3062 - val_loss: 0.3635; rmse: 0.5606 - rmse_val: 0.6029;  r2: 0.9446 - r2_val: 0.9214                                                                                                    \n",
      "epoch: 0034, loss: 0.2799 - val_loss: 0.4768; rmse: 0.6484 - rmse_val: 0.6905;  r2: 0.9441 - r2_val: 0.9176                                                                                                    \n",
      "epoch: 0035, loss: 0.2931 - val_loss: 0.3305; rmse: 0.5063 - rmse_val: 0.5749;  r2: 0.9466 - r2_val: 0.9227                                                                                                    \n",
      "epoch: 0036, loss: 0.3006 - val_loss: 0.4047; rmse: 0.6043 - rmse_val: 0.6361;  r2: 0.9461 - r2_val: 0.9224                                                                                                    \n",
      "epoch: 0037, loss: 0.2981 - val_loss: 0.3243; rmse: 0.4867 - rmse_val: 0.5695;  r2: 0.9492 - r2_val: 0.9286                                                                                                    \n",
      "epoch: 0038, loss: 0.2726 - val_loss: 0.3323; rmse: 0.4875 - rmse_val: 0.5765;  r2: 0.9514 - r2_val: 0.9299                                                                                                    \n",
      "epoch: 0039, loss: 0.2551 - val_loss: 0.2991; rmse: 0.4756 - rmse_val: 0.5469;  r2: 0.9492 - r2_val: 0.9282                                                                                                    \n",
      "epoch: 0040, loss: 0.2620 - val_loss: 0.3092; rmse: 0.4622 - rmse_val: 0.5560;  r2: 0.9535 - r2_val: 0.9284                                                                                                    \n",
      "epoch: 0041, loss: 0.2869 - val_loss: 0.3392; rmse: 0.4776 - rmse_val: 0.5824;  r2: 0.9494 - r2_val: 0.9192                                                                                                    \n",
      "epoch: 0042, loss: 0.2662 - val_loss: 0.4270; rmse: 0.5414 - rmse_val: 0.6535;  r2: 0.9535 - r2_val: 0.9284                                                                                                    \n",
      "epoch: 0043, loss: 0.2397 - val_loss: 0.3114; rmse: 0.4622 - rmse_val: 0.5580;  r2: 0.9561 - r2_val: 0.9286                                                                                                    \n",
      "epoch: 0044, loss: 0.2671 - val_loss: 0.3566; rmse: 0.4845 - rmse_val: 0.5972;  r2: 0.9554 - r2_val: 0.9291                                                                                                    \n",
      "epoch: 0045, loss: 0.2374 - val_loss: 0.3130; rmse: 0.4504 - rmse_val: 0.5594;  r2: 0.9587 - r2_val: 0.9270                                                                                                    \n",
      "epoch: 0046, loss: 0.2272 - val_loss: 0.3308; rmse: 0.4616 - rmse_val: 0.5751;  r2: 0.9586 - r2_val: 0.9323                                                                                                    \n",
      "epoch: 0047, loss: 0.2338 - val_loss: 0.3269; rmse: 0.4340 - rmse_val: 0.5718;  r2: 0.9593 - r2_val: 0.9264                                                                                                    \n",
      "epoch: 0048, loss: 0.2223 - val_loss: 0.3208; rmse: 0.4456 - rmse_val: 0.5664;  r2: 0.9598 - r2_val: 0.9328                                                                                                    \n",
      "epoch: 0049, loss: 0.2518 - val_loss: 0.2876; rmse: 0.4091 - rmse_val: 0.5363;  r2: 0.9630 - r2_val: 0.9312                                                                                                    \n",
      "epoch: 0050, loss: 0.2403 - val_loss: 0.4304; rmse: 0.5580 - rmse_val: 0.6561;  r2: 0.9593 - r2_val: 0.9340                                                                                                    \n",
      "epoch: 0051, loss: 0.2387 - val_loss: 0.3903; rmse: 0.4914 - rmse_val: 0.6248;  r2: 0.9599 - r2_val: 0.9287                                                                                                    \n",
      "epoch: 0052, loss: 0.2165 - val_loss: 0.3264; rmse: 0.4354 - rmse_val: 0.5713;  r2: 0.9621 - r2_val: 0.9302                                                                                                    \n",
      "epoch: 0053, loss: 0.2247 - val_loss: 0.3216; rmse: 0.4137 - rmse_val: 0.5671;  r2: 0.9630 - r2_val: 0.9242                                                                                                    \n",
      "epoch: 0054, loss: 0.2160 - val_loss: 0.3178; rmse: 0.4315 - rmse_val: 0.5637;  r2: 0.9644 - r2_val: 0.9322                                                                                                    \n",
      "epoch: 0055, loss: 0.2004 - val_loss: 0.3023; rmse: 0.4095 - rmse_val: 0.5498;  r2: 0.9636 - r2_val: 0.9281                                                                                                    \n",
      "epoch: 0056, loss: 0.1904 - val_loss: 0.4416; rmse: 0.5544 - rmse_val: 0.6645;  r2: 0.9642 - r2_val: 0.9370                                                                                                    \n",
      "epoch: 0057, loss: 0.2299 - val_loss: 0.3148; rmse: 0.4093 - rmse_val: 0.5611;  r2: 0.9661 - r2_val: 0.9275                                                                                                    \n",
      "epoch: 0058, loss: 0.1851 - val_loss: 0.2924; rmse: 0.3774 - rmse_val: 0.5408;  r2: 0.9680 - r2_val: 0.9301                                                                                                    \n",
      "epoch: 0059, loss: 0.2198 - val_loss: 0.3244; rmse: 0.4567 - rmse_val: 0.5695;  r2: 0.9662 - r2_val: 0.9306                                                                                                    \n",
      "epoch: 0060, loss: 0.1877 - val_loss: 0.2889; rmse: 0.3750 - rmse_val: 0.5375;  r2: 0.9683 - r2_val: 0.9308                                                                                                    \n",
      "epoch: 0061, loss: 0.1932 - val_loss: 0.3415; rmse: 0.4429 - rmse_val: 0.5844;  r2: 0.9677 - r2_val: 0.9350                                                                                                    \n",
      "epoch: 0062, loss: 0.1709 - val_loss: 0.3098; rmse: 0.4139 - rmse_val: 0.5566;  r2: 0.9649 - r2_val: 0.9286                                                                                                    \n",
      "epoch: 0063, loss: 0.1877 - val_loss: 0.2994; rmse: 0.3873 - rmse_val: 0.5472;  r2: 0.9669 - r2_val: 0.9286                                                                                                    \n",
      "epoch: 0064, loss: 0.1789 - val_loss: 0.3387; rmse: 0.4333 - rmse_val: 0.5819;  r2: 0.9710 - r2_val: 0.9278                                                                                                    \n",
      "epoch: 0065, loss: 0.1751 - val_loss: 0.4056; rmse: 0.5355 - rmse_val: 0.6369;  r2: 0.9717 - r2_val: 0.9360                                                                                                    \n",
      "epoch: 0066, loss: 0.1957 - val_loss: 0.3705; rmse: 0.4649 - rmse_val: 0.6087;  r2: 0.9703 - r2_val: 0.9355                                                                                                    \n",
      "epoch: 0067, loss: 0.2040 - val_loss: 0.4259; rmse: 0.5310 - rmse_val: 0.6526;  r2: 0.9711 - r2_val: 0.9314                                                                                                    \n",
      "epoch: 0068, loss: 0.2111 - val_loss: 0.3597; rmse: 0.4213 - rmse_val: 0.5997;  r2: 0.9701 - r2_val: 0.9297                                                                                                    \n",
      "epoch: 0069, loss: 0.1720 - val_loss: 0.2820; rmse: 0.3765 - rmse_val: 0.5310;  r2: 0.9706 - r2_val: 0.9327                                                                                                    \n",
      "epoch: 0070, loss: 0.1624 - val_loss: 0.3706; rmse: 0.4257 - rmse_val: 0.6087;  r2: 0.9740 - r2_val: 0.9317                                                                                                    \n",
      "epoch: 0071, loss: 0.1691 - val_loss: 0.3328; rmse: 0.3942 - rmse_val: 0.5769;  r2: 0.9722 - r2_val: 0.9339                                                                                                    \n",
      "epoch: 0072, loss: 0.1691 - val_loss: 0.2873; rmse: 0.3493 - rmse_val: 0.5360;  r2: 0.9729 - r2_val: 0.9314                                                                                                    \n",
      "epoch: 0073, loss: 0.1702 - val_loss: 0.2982; rmse: 0.3739 - rmse_val: 0.5461;  r2: 0.9735 - r2_val: 0.9356                                                                                                    \n",
      "epoch: 0074, loss: 0.1709 - val_loss: 0.3431; rmse: 0.4179 - rmse_val: 0.5857;  r2: 0.9686 - r2_val: 0.9225                                                                                                    \n",
      "epoch: 0075, loss: 0.1854 - val_loss: 0.2839; rmse: 0.3505 - rmse_val: 0.5329;  r2: 0.9725 - r2_val: 0.9320                                                                                                    \n",
      "epoch: 0076, loss: 0.1602 - val_loss: 0.3159; rmse: 0.3855 - rmse_val: 0.5620;  r2: 0.9739 - r2_val: 0.9360                                                                                                    \n",
      "epoch: 0077, loss: 0.1582 - val_loss: 0.2699; rmse: 0.3440 - rmse_val: 0.5196;  r2: 0.9744 - r2_val: 0.9355                                                                                                    \n",
      "epoch: 0078, loss: 0.1429 - val_loss: 0.3023; rmse: 0.3874 - rmse_val: 0.5498;  r2: 0.9712 - r2_val: 0.9316                                                                                                    \n",
      "epoch: 0079, loss: 0.1527 - val_loss: 0.2956; rmse: 0.3280 - rmse_val: 0.5437;  r2: 0.9760 - r2_val: 0.9301                                                                                                    \n",
      "epoch: 0080, loss: 0.1504 - val_loss: 0.3049; rmse: 0.3796 - rmse_val: 0.5521;  r2: 0.9767 - r2_val: 0.9375                                                                                                    \n",
      "epoch: 0081, loss: 0.1562 - val_loss: 0.2844; rmse: 0.3474 - rmse_val: 0.5333;  r2: 0.9760 - r2_val: 0.9340                                                                                                    \n",
      "epoch: 0082, loss: 0.1515 - val_loss: 0.3071; rmse: 0.3865 - rmse_val: 0.5542;  r2: 0.9760 - r2_val: 0.9336                                                                                                    \n",
      "epoch: 0083, loss: 0.1362 - val_loss: 0.2943; rmse: 0.3486 - rmse_val: 0.5425;  r2: 0.9743 - r2_val: 0.9313                                                                                                    \n",
      "epoch: 0084, loss: 0.1547 - val_loss: 0.4175; rmse: 0.4904 - rmse_val: 0.6461;  r2: 0.9753 - r2_val: 0.9355                                                                                                    \n",
      "epoch: 0085, loss: 0.1407 - val_loss: 0.3147; rmse: 0.3597 - rmse_val: 0.5610;  r2: 0.9788 - r2_val: 0.9315                                                                                                    \n",
      "epoch: 0086, loss: 0.1444 - val_loss: 0.2950; rmse: 0.3252 - rmse_val: 0.5431;  r2: 0.9780 - r2_val: 0.9305                                                                                                    \n",
      "epoch: 0087, loss: 0.1359 - val_loss: 0.2817; rmse: 0.3174 - rmse_val: 0.5307;  r2: 0.9779 - r2_val: 0.9328                                                                                                    \n",
      "epoch: 0088, loss: 0.1322 - val_loss: 0.3170; rmse: 0.3996 - rmse_val: 0.5630;  r2: 0.9729 - r2_val: 0.9345                                                                                                    \n",
      "epoch: 0089, loss: 0.1371 - val_loss: 0.2690; rmse: 0.3146 - rmse_val: 0.5187;  r2: 0.9789 - r2_val: 0.9364                                                                                                    \n",
      "epoch: 0090, loss: 0.1465 - val_loss: 0.2960; rmse: 0.3449 - rmse_val: 0.5440;  r2: 0.9770 - r2_val: 0.9329                                                                                                    \n",
      "epoch: 0091, loss: 0.1333 - val_loss: 0.3027; rmse: 0.3552 - rmse_val: 0.5502;  r2: 0.9796 - r2_val: 0.9324                                                                                                    \n",
      "epoch: 0092, loss: 0.1237 - val_loss: 0.2892; rmse: 0.3191 - rmse_val: 0.5378;  r2: 0.9778 - r2_val: 0.9311                                                                                                    \n",
      "epoch: 0093, loss: 0.1544 - val_loss: 0.3303; rmse: 0.3830 - rmse_val: 0.5748;  r2: 0.9773 - r2_val: 0.9286                                                                                                    \n",
      "epoch: 0094, loss: 0.1449 - val_loss: 0.3128; rmse: 0.3578 - rmse_val: 0.5592;  r2: 0.9791 - r2_val: 0.9305                                                                                                    \n",
      "epoch: 0095, loss: 0.1319 - val_loss: 0.4002; rmse: 0.4946 - rmse_val: 0.6327;  r2: 0.9792 - r2_val: 0.9362                                                                                                    \n",
      "epoch: 0096, loss: 0.1381 - val_loss: 0.2924; rmse: 0.3189 - rmse_val: 0.5407;  r2: 0.9785 - r2_val: 0.9339                                                                                                    \n",
      "epoch: 0097, loss: 0.1391 - val_loss: 0.3491; rmse: 0.4372 - rmse_val: 0.5909;  r2: 0.9777 - r2_val: 0.9344                                                                                                    \n",
      "epoch: 0098, loss: 0.1304 - val_loss: 0.2897; rmse: 0.3073 - rmse_val: 0.5382;  r2: 0.9818 - r2_val: 0.9322                                                                                                    \n",
      "epoch: 0099, loss: 0.1195 - val_loss: 0.3039; rmse: 0.3551 - rmse_val: 0.5513;  r2: 0.9775 - r2_val: 0.9335                                                                                                    \n",
      "epoch: 0100, loss: 0.1126 - val_loss: 0.2657; rmse: 0.2792 - rmse_val: 0.5154;  r2: 0.9827 - r2_val: 0.9364                                                                                                    \n",
      "epoch: 0101, loss: 0.1203 - val_loss: 0.2883; rmse: 0.2996 - rmse_val: 0.5370;  r2: 0.9819 - r2_val: 0.9320                                                                                                    \n",
      "epoch: 0102, loss: 0.1100 - val_loss: 0.2836; rmse: 0.2919 - rmse_val: 0.5326;  r2: 0.9818 - r2_val: 0.9340                                                                                                    \n",
      "epoch: 0103, loss: 0.1128 - val_loss: 0.3431; rmse: 0.4173 - rmse_val: 0.5857;  r2: 0.9808 - r2_val: 0.9351                                                                                                    \n",
      "epoch: 0104, loss: 0.1178 - val_loss: 0.3171; rmse: 0.3462 - rmse_val: 0.5631;  r2: 0.9824 - r2_val: 0.9318                                                                                                    \n",
      "epoch: 0105, loss: 0.1085 - val_loss: 0.3048; rmse: 0.3096 - rmse_val: 0.5521;  r2: 0.9800 - r2_val: 0.9289                                                                                                    \n",
      "epoch: 0106, loss: 0.1113 - val_loss: 0.2627; rmse: 0.2962 - rmse_val: 0.5125;  r2: 0.9833 - r2_val: 0.9382                                                                                                    \n",
      "epoch: 0107, loss: 0.1086 - val_loss: 0.2746; rmse: 0.2836 - rmse_val: 0.5240;  r2: 0.9820 - r2_val: 0.9342                                                                                                    \n",
      "epoch: 0108, loss: 0.1112 - val_loss: 0.2636; rmse: 0.2963 - rmse_val: 0.5134;  r2: 0.9807 - r2_val: 0.9368                                                                                                    \n",
      "epoch: 0109, loss: 0.1145 - val_loss: 0.2717; rmse: 0.2839 - rmse_val: 0.5213;  r2: 0.9829 - r2_val: 0.9379                                                                                                    \n",
      "epoch: 0110, loss: 0.1092 - val_loss: 0.2804; rmse: 0.2882 - rmse_val: 0.5295;  r2: 0.9832 - r2_val: 0.9362                                                                                                    \n",
      "epoch: 0111, loss: 0.1460 - val_loss: 0.3484; rmse: 0.3646 - rmse_val: 0.5903;  r2: 0.9788 - r2_val: 0.9326                                                                                                    \n",
      "epoch: 0112, loss: 0.1049 - val_loss: 0.2794; rmse: 0.2944 - rmse_val: 0.5286;  r2: 0.9836 - r2_val: 0.9379                                                                                                    \n",
      "epoch: 0113, loss: 0.1175 - val_loss: 0.2742; rmse: 0.2725 - rmse_val: 0.5236;  r2: 0.9852 - r2_val: 0.9347                                                                                                    \n",
      "epoch: 0114, loss: 0.1295 - val_loss: 0.2961; rmse: 0.3738 - rmse_val: 0.5442;  r2: 0.9795 - r2_val: 0.9373                                                                                                    \n",
      "epoch: 0115, loss: 0.1233 - val_loss: 0.3135; rmse: 0.3578 - rmse_val: 0.5599;  r2: 0.9793 - r2_val: 0.9338                                                                                                    \n",
      "epoch: 0116, loss: 0.1014 - val_loss: 0.3336; rmse: 0.3999 - rmse_val: 0.5776;  r2: 0.9786 - r2_val: 0.9294                                                                                                    \n",
      "epoch: 0117, loss: 0.1146 - val_loss: 0.3264; rmse: 0.3242 - rmse_val: 0.5713;  r2: 0.9815 - r2_val: 0.9290                                                                                                    \n",
      "epoch: 0118, loss: 0.0951 - val_loss: 0.3248; rmse: 0.3560 - rmse_val: 0.5699;  r2: 0.9836 - r2_val: 0.9330                                                                                                    \n",
      "epoch: 0119, loss: 0.1045 - val_loss: 0.2622; rmse: 0.2744 - rmse_val: 0.5120;  r2: 0.9848 - r2_val: 0.9380                                                                                                    \n",
      "epoch: 0120, loss: 0.0951 - val_loss: 0.3263; rmse: 0.3258 - rmse_val: 0.5712;  r2: 0.9828 - r2_val: 0.9343                                                                                                    \n",
      "epoch: 0121, loss: 0.1115 - val_loss: 0.3007; rmse: 0.2988 - rmse_val: 0.5484;  r2: 0.9852 - r2_val: 0.9355                                                                                                    \n",
      "epoch: 0122, loss: 0.0972 - val_loss: 0.2757; rmse: 0.2926 - rmse_val: 0.5250;  r2: 0.9842 - r2_val: 0.9404                                                                                                    \n",
      "epoch: 0123, loss: 0.0926 - val_loss: 0.3080; rmse: 0.2979 - rmse_val: 0.5549;  r2: 0.9850 - r2_val: 0.9347                                                                                                    \n",
      "epoch: 0124, loss: 0.1141 - val_loss: 0.2630; rmse: 0.3104 - rmse_val: 0.5128;  r2: 0.9798 - r2_val: 0.9376                                                                                                    \n",
      "epoch: 0125, loss: 0.1026 - val_loss: 0.3522; rmse: 0.3602 - rmse_val: 0.5934;  r2: 0.9850 - r2_val: 0.9349                                                                                                    \n",
      "epoch: 0126, loss: 0.0853 - val_loss: 0.2574; rmse: 0.2601 - rmse_val: 0.5073;  r2: 0.9867 - r2_val: 0.9412                                                                                                    \n",
      "epoch: 0127, loss: 0.0880 - val_loss: 0.2617; rmse: 0.2435 - rmse_val: 0.5116;  r2: 0.9871 - r2_val: 0.9389                                                                                                    \n",
      "epoch: 0128, loss: 0.1150 - val_loss: 0.3186; rmse: 0.3396 - rmse_val: 0.5644;  r2: 0.9826 - r2_val: 0.9303                                                                                                    \n",
      "epoch: 0129, loss: 0.1035 - val_loss: 0.2830; rmse: 0.2766 - rmse_val: 0.5319;  r2: 0.9869 - r2_val: 0.9385                                                                                                    \n",
      "epoch: 0130, loss: 0.1012 - val_loss: 0.2882; rmse: 0.2612 - rmse_val: 0.5368;  r2: 0.9860 - r2_val: 0.9325                                                                                                    \n",
      "epoch: 0131, loss: 0.0951 - val_loss: 0.2344; rmse: 0.2519 - rmse_val: 0.4842;  r2: 0.9863 - r2_val: 0.9437                                                                                                    \n",
      "epoch: 0132, loss: 0.1385 - val_loss: 0.2602; rmse: 0.2768 - rmse_val: 0.5101;  r2: 0.9846 - r2_val: 0.9381                                                                                                    \n",
      "epoch: 0133, loss: 0.0893 - val_loss: 0.2704; rmse: 0.2945 - rmse_val: 0.5200;  r2: 0.9861 - r2_val: 0.9381                                                                                                    \n",
      "epoch: 0134, loss: 0.0942 - val_loss: 0.3148; rmse: 0.3436 - rmse_val: 0.5611;  r2: 0.9833 - r2_val: 0.9349                                                                                                    \n",
      "epoch: 0135, loss: 0.0931 - val_loss: 0.2722; rmse: 0.2526 - rmse_val: 0.5217;  r2: 0.9858 - r2_val: 0.9351                                                                                                    \n",
      "epoch: 0136, loss: 0.0814 - val_loss: 0.2781; rmse: 0.2486 - rmse_val: 0.5274;  r2: 0.9872 - r2_val: 0.9358                                                                                                    \n",
      "epoch: 0137, loss: 0.1081 - val_loss: 0.2790; rmse: 0.2627 - rmse_val: 0.5282;  r2: 0.9847 - r2_val: 0.9338                                                                                                    \n",
      "epoch: 0138, loss: 0.0871 - val_loss: 0.2591; rmse: 0.2553 - rmse_val: 0.5090;  r2: 0.9857 - r2_val: 0.9381                                                                                                    \n",
      "epoch: 0139, loss: 0.0859 - val_loss: 0.2510; rmse: 0.2309 - rmse_val: 0.5010;  r2: 0.9882 - r2_val: 0.9398                                                                                                    \n",
      "epoch: 0140, loss: 0.0990 - val_loss: 0.2762; rmse: 0.2803 - rmse_val: 0.5256;  r2: 0.9862 - r2_val: 0.9374                                                                                                    \n",
      "epoch: 0141, loss: 0.0870 - val_loss: 0.2982; rmse: 0.3653 - rmse_val: 0.5460;  r2: 0.9841 - r2_val: 0.9411                                                                                                    \n",
      "epoch: 0142, loss: 0.1082 - val_loss: 0.2635; rmse: 0.2626 - rmse_val: 0.5133;  r2: 0.9846 - r2_val: 0.9373                                                                                                    \n",
      "epoch: 0143, loss: 0.0746 - val_loss: 0.2851; rmse: 0.2707 - rmse_val: 0.5340;  r2: 0.9878 - r2_val: 0.9349                                                                                                    \n",
      "epoch: 0144, loss: 0.0957 - val_loss: 0.2998; rmse: 0.2471 - rmse_val: 0.5476;  r2: 0.9875 - r2_val: 0.9319                                                                                                    \n",
      "epoch: 0145, loss: 0.0740 - val_loss: 0.2525; rmse: 0.2337 - rmse_val: 0.5025;  r2: 0.9890 - r2_val: 0.9403                                                                                                    \n",
      "epoch: 0146, loss: 0.0842 - val_loss: 0.2608; rmse: 0.2490 - rmse_val: 0.5107;  r2: 0.9875 - r2_val: 0.9391                                                                                                    \n",
      "epoch: 0147, loss: 0.0758 - val_loss: 0.2612; rmse: 0.2966 - rmse_val: 0.5111;  r2: 0.9852 - r2_val: 0.9395                                                                                                    \n",
      "epoch: 0148, loss: 0.0771 - val_loss: 0.2550; rmse: 0.2294 - rmse_val: 0.5049;  r2: 0.9883 - r2_val: 0.9390                                                                                                    \n",
      "epoch: 0149, loss: 0.0794 - val_loss: 0.2736; rmse: 0.2376 - rmse_val: 0.5231;  r2: 0.9877 - r2_val: 0.9346                                                                                                    \n",
      "epoch: 0150, loss: 0.0876 - val_loss: 0.2934; rmse: 0.2810 - rmse_val: 0.5416;  r2: 0.9882 - r2_val: 0.9346                                                                                                    \n",
      "epoch: 0151, loss: 0.0867 - val_loss: 0.2638; rmse: 0.2568 - rmse_val: 0.5136;  r2: 0.9885 - r2_val: 0.9396                                                                                                    \n",
      "\n",
      "Restoring model weights from the end of the best epoch.\n",
      "\n",
      "Epoch 00151: early stopping\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "RegressionEstimator(batch_size=8, gpuid='1', patience=20)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fit your model\n",
    "clf.fit(trainX, trainY, validX, validY)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(clf.history.history).plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best epochs: 130.00, Best MSE: 0.23\n"
     ]
    }
   ],
   "source": [
    "print('Best epochs: %.2f, Best MSE: %.2f' % (clf._performance.best_epoch, clf._performance.best))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "testY_pred = clf.predict(testX)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f06b8282978>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(testY, testY_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([0.5598364810998153], [0.932323885893874])"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rmse, r2 = clf._performance.evaluate(testX, testY)\n",
    "rmse, r2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.932323885893874"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.score(testX, testY)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([0.4841993583649496], [0.9436707998060943])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf._performance.evaluate(validX, validY)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}